diff --git a/.github/workflows/create-version-tag.yml b/.github/workflows/create-version-tag.yml
index 8be17b4a3..630caef92 100644
--- a/.github/workflows/create-version-tag.yml
+++ b/.github/workflows/create-version-tag.yml
@@ -23,7 +23,7 @@ jobs:
       - name: Build
         uses: gradle/gradle-build-action@v3
         with:
-          arguments: :xef-openai-client-generator:openaiClientGenerate build
+          arguments: build
 
       - name: Read version
         id: read-version
diff --git a/.github/workflows/openai-api-check.yml b/.github/workflows/openai-api-check.yml
index 2bb319f84..9db55909d 100644
--- a/.github/workflows/openai-api-check.yml
+++ b/.github/workflows/openai-api-check.yml
@@ -50,7 +50,7 @@ jobs:
             const octokit = new Octokit();
             const owner = ''xebia-functional'';
             const repo = ''xef'';
-            const filePath = ''openai-client/generator/config/openai-api-commit'';
+            const filePath = ''core/openai-api-commit'';
             const branch = ''main'';
             const response = await octokit.repos.getContent({
               owner,
@@ -78,7 +78,7 @@ jobs:
             const octokit = new Octokit();
             const owner = ''xebia-functional'';
             const repo = ''xef'';
-            const filePath = ''openai-client/generator/config/openai-api-commit'';
+            const filePath = ''core/openai-api-commit'';
             const existing = await octokit.pulls.list({
               owner,
               repo,
@@ -121,7 +121,7 @@ jobs:
       - name: Update commit
         if: env.CURRENT_API_COMMIT != env.LATEST_API_COMMIT
         run: |
-          echo $LATEST_API_COMMIT > $GITHUB_WORKSPACE/openai-client/generator/config/openai-api-commit
+          echo $LATEST_API_COMMIT > $GITHUB_WORKSPACE/core/openai-api-commit
       - name: Set up Java
         if: env.CURRENT_API_COMMIT != env.LATEST_API_COMMIT
         uses: actions/setup-java@v4
diff --git a/.github/workflows/publish-development-version.yml b/.github/workflows/publish-development-version.yml
index 1d0aa7895..7406faab7 100644
--- a/.github/workflows/publish-development-version.yml
+++ b/.github/workflows/publish-development-version.yml
@@ -34,7 +34,7 @@ jobs:
       - name: Assemble
         uses: gradle/gradle-build-action@v3
         with:
-          arguments: :xef-openai-client-generator:openaiClientGenerate assemble
+          arguments: assemble
 
       - name: Upload reports
         if: failure()
diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml
index b324b6eb4..9498adf8e 100644
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@@ -32,7 +32,7 @@ jobs:
       - name: Assemble
         uses: gradle/gradle-build-action@v3
         with:
-          arguments: :xef-openai-client-generator:openaiClientGenerate assemble
+          arguments: assemble
 
       - name: Upload reports
         if: failure()
diff --git a/build.gradle.kts b/build.gradle.kts
index 0dc19a5bd..594022ff9 100644
--- a/build.gradle.kts
+++ b/build.gradle.kts
@@ -28,12 +28,10 @@ fun Project.configureBuildAndTestTask(taskName: String, moduleType: ModulePlatfo
     doLast {
       when (moduleType) {
         ModulePlatformType.SINGLE -> {
-          project.exec { commandLine(gradleCommand, ":xef-openai-client-generator:openaiClientGenerate") }
           val excludedModules = includeOrNotModulesToCommand(multiPlatformModules, platform, false)
           project.exec { commandLine(gradleCommand, "build", *excludedModules) }
         }
         ModulePlatformType.MULTI -> {
-          project.exec { commandLine(gradleCommand, ":xef-openai-client-generator:openaiClientGenerate") }
           val includedModules = includeOrNotModulesToCommand(multiPlatformModules, platform, true)
           project.exec { commandLine(gradleCommand, *includedModules) }
         }
@@ -65,3 +63,7 @@ fun getGradleCommand(platform: String): String {
 configureBuildAndTestTask("buildAndTestMultip", ModulePlatformType.MULTI)
 configureBuildAndTestTask("buildAndTestSinglep", ModulePlatformType.SINGLE)
 
+tasks.register<DownloadOpenAI>("downloadOpenAIAPI") {
+    input.set(file("core/openai-api-commit"))
+    output.set(file("core/openai-api.yaml"))
+}
diff --git a/buildSrc/src/main/kotlin/DownloadOpenAI.kt b/buildSrc/src/main/kotlin/DownloadOpenAI.kt
new file mode 100644
index 000000000..09900038f
--- /dev/null
+++ b/buildSrc/src/main/kotlin/DownloadOpenAI.kt
@@ -0,0 +1,45 @@
+import org.gradle.api.DefaultTask
+import org.gradle.api.file.RegularFileProperty
+import org.gradle.api.tasks.*
+import org.gradle.api.tasks.options.Option
+import java.net.URL
+import java.nio.channels.Channels
+
+@CacheableTask
+abstract class DownloadOpenAI : DefaultTask() {
+    init {
+        description = "Download the OpenAI API client"
+        group = "openai"
+    }
+
+    @get:InputFile
+    @get:Option(description = "Commit hash of the fetched OpenAI OpenAPI Yaml file")
+    @get:PathSensitive(PathSensitivity.RELATIVE)
+    abstract val input: RegularFileProperty
+
+    @get:OutputFile
+    @get:Option(description = "Path of file to store the fetched OpenAI OpenAPI Yaml")
+    abstract val output: RegularFileProperty
+
+    @TaskAction
+    fun download() {
+        val commit = readCommit()
+        downloadAPI(commit)
+    }
+
+    private fun readCommit(): String {
+        val asFile = input.get().asFile
+        val readText = asFile.readText()
+        require(readText.isNotEmpty()) { "${asFile.path} Commit hash is empty" }
+        return readText.trim { it <= ' ' }
+    }
+
+    private fun downloadAPI(commit: String) {
+        val url = URL("https://raw.githubusercontent.com/openai/openai-openapi/$commit/openapi.yaml")
+        Channels.newChannel(url.openStream()).use { channel ->
+            output.asFile.get().outputStream().use { output ->
+                output.channel.transferFrom(channel, 0, Long.MAX_VALUE)
+            }
+        }
+    }
+}
diff --git a/core/build.gradle.kts b/core/build.gradle.kts
index de1530459..d7580df54 100644
--- a/core/build.gradle.kts
+++ b/core/build.gradle.kts
@@ -16,6 +16,13 @@ plugins {
     alias(libs.plugins.arrow.gradle.publish)
     alias(libs.plugins.semver.gradle)
     alias(libs.plugins.detekt)
+    alias(libs.plugins.openapi.kt)
+}
+
+openApiConfig {
+    spec("OpenAI", file("openai-api.yaml")) {
+        packageName = "com.xebia.functional.xef.openapi"
+    }
 }
 
 dependencies {
@@ -47,11 +54,11 @@ kotlin {
                     classpath = compileDependencyFiles + runtimeDependencyFiles + output.allOutputs
                     testClassesDirs = output.classesDirs
                     testLogging { events("passed") }
-              }
-          }
-          val test by compilations.getting
-          integrationTest.associateWith(test)
-      }
+                }
+            }
+            val test by compilations.getting
+            integrationTest.associateWith(test)
+        }
     }
     js(IR) {
         browser()
@@ -62,73 +69,72 @@ kotlin {
     macosArm64()
     mingwX64()
     sourceSets {
-      all {
-          languageSettings.optIn("kotlin.ExperimentalStdlibApi")
-      }
-      val commonMain by getting {
-          dependencies {
-              api(libs.bundles.arrow)
-              api(libs.kotlinx.serialization.json)
-              api(libs.kotlinx.serialization.yaml)
-              api(libs.ktor.utils)
-              api(projects.xefTokenizer)
-              api(projects.xefOpenaiClient)
-              implementation(libs.bundles.ktor.client)
-              implementation(libs.klogging)
-              implementation(libs.uuid)
-              implementation(libs.ktor.client.logging)
-              implementation(libs.klogging)
-          }
-      }
-      val commonTest by getting {
-          dependencies {
-              implementation(libs.kotest.property)
-              implementation(libs.kotest.framework)
-              implementation(libs.kotest.assertions)
-          }
-      }
-      val jvmMain by getting {
-          dependencies {
-              implementation(libs.ktor.http)
-              implementation(libs.rss.reader)
-              api(libs.jackson)
-              api(libs.jackson.schema)
-              api(libs.jackson.schema.jakarta)
-              api(libs.jakarta.validation)
-              implementation(libs.kotlinx.coroutines.reactive)
-              api(libs.ktor.client.cio)
-          }
-      }
-      val jsMain by getting {
-          dependencies {
-              api(libs.ktor.client.js)
-          }
-      }
-      val jvmTest by getting {
-          dependencies {
-              implementation(libs.kotest.junit5)
-          }
-      }
-      val linuxX64Main by getting {
-          dependencies {
-              implementation(libs.ktor.client.cio)
-          }
-      }
-      val macosX64Main by getting {
-          dependencies {
-              implementation(libs.ktor.client.cio)
-          }
-      }
-      val macosArm64Main by getting {
-          dependencies {
-              implementation(libs.ktor.client.cio)
-          }
-      }
-      val mingwX64Main by getting {
-          dependencies {
-              implementation(libs.ktor.client.winhttp)
-          }
-      }
+        all {
+            languageSettings.optIn("kotlin.ExperimentalStdlibApi")
+        }
+        val commonMain by getting {
+            dependencies {
+                api(libs.bundles.arrow)
+                api(libs.kotlinx.serialization.json)
+                api(libs.kotlinx.serialization.yaml)
+                api(libs.ktor.utils)
+                api(projects.xefTokenizer)
+                implementation(libs.bundles.ktor.client)
+                implementation(libs.klogging)
+                implementation(libs.uuid)
+                implementation(libs.ktor.client.logging)
+                implementation(libs.klogging)
+            }
+        }
+        val commonTest by getting {
+            dependencies {
+                implementation(libs.kotest.property)
+                implementation(libs.kotest.framework)
+                implementation(libs.kotest.assertions)
+            }
+        }
+        val jvmMain by getting {
+            dependencies {
+                implementation(libs.ktor.http)
+                implementation(libs.rss.reader)
+                api(libs.jackson)
+                api(libs.jackson.schema)
+                api(libs.jackson.schema.jakarta)
+                api(libs.jakarta.validation)
+                implementation(libs.kotlinx.coroutines.reactive)
+                api(libs.ktor.client.cio)
+            }
+        }
+        val jsMain by getting {
+            dependencies {
+                api(libs.ktor.client.js)
+            }
+        }
+        val jvmTest by getting {
+            dependencies {
+                implementation(libs.kotest.junit5)
+            }
+        }
+        val linuxX64Main by getting {
+            dependencies {
+                implementation(libs.ktor.client.cio)
+            }
+        }
+        val macosX64Main by getting {
+            dependencies {
+                implementation(libs.ktor.client.cio)
+            }
+        }
+        val macosArm64Main by getting {
+            dependencies {
+                implementation(libs.ktor.client.cio)
+            }
+        }
+        val mingwX64Main by getting {
+            dependencies {
+                implementation(libs.ktor.client.winhttp)
+            }
+        }
     }
 }
 
diff --git a/core/openai-api-commit b/core/openai-api-commit
new file mode 100644
index 000000000..4746a6be6
--- /dev/null
+++ b/core/openai-api-commit
@@ -0,0 +1 @@
+df5699f5c48fe455d9d036198e77fdb637edee29
\ No newline at end of file
diff --git a/openai-client/generator/config/openai-api.yaml b/core/openai-api.yaml
similarity index 100%
rename from openai-client/generator/config/openai-api.yaml
rename to core/openai-api.yaml
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/AI.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/AI.kt
index b665a76aa..c3a44a099 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/AI.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/AI.kt
@@ -1,13 +1,15 @@
 package com.xebia.functional.xef
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequest
 import com.xebia.functional.xef.conversation.AiDsl
 import com.xebia.functional.xef.llm.models.modelType
 import com.xebia.functional.xef.llm.prompt
 import com.xebia.functional.xef.llm.promptStreaming
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import com.xebia.functional.xef.prompt.Prompt
 import kotlinx.coroutines.flow.Flow
 import kotlinx.coroutines.flow.channelFlow
+import kotlinx.serialization.json.JsonObject
+import kotlinx.serialization.json.JsonPrimitive
 
 class AI<out A>(private val config: AIConfig, val serializer: Tool<A>) {
 
@@ -59,17 +61,19 @@ class AI<out A>(private val config: AIConfig, val serializer: Tool<A>) {
     val encoding = prompt.model.modelType(forFunctions = false).encoding
     val cases = serializer.cases
     val logitBias =
-      cases
-        .flatMap {
-          val result = encoding.encode(it.function.name)
-          if (result.size > 1) {
-            error("Cannot encode enum case $it into one token")
+      JsonObject(
+        cases
+          .flatMap {
+            val result = encoding.encode(it.function.name)
+            if (result.size > 1) {
+              error("Cannot encode enum case $it into one token")
+            }
+            result
           }
-          result
-        }
-        .associate { "$it" to 100 }
+          .associate { "$it" to JsonPrimitive(100) }
+      )
     val result =
-      config.api.createChatCompletion(
+      config.api.completions.createChatCompletion(
         CreateChatCompletionRequest(
           messages = prompt.messages,
           model = prompt.model,
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/AIConfig.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/AIConfig.kt
index a9d0f945f..a44be8b9a 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/AIConfig.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/AIConfig.kt
@@ -1,13 +1,13 @@
 package com.xebia.functional.xef
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.api.OpenAI
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.conversation.Conversation
+import com.xebia.functional.xef.openapi.Chat
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.OpenAI
 
 data class AIConfig(
   val tools: List<Tool<*>> = emptyList(),
-  val model: CreateChatCompletionRequestModel = CreateChatCompletionRequestModel.gpt_4o,
+  val model: CreateChatCompletionRequest.Model = CreateChatCompletionRequest.Model.Gpt4o,
   val config: Config = Config(),
   val openAI: OpenAI = OpenAI(config, logRequests = false),
   val api: Chat = openAI.chat,
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/Config.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/Config.kt
index 2f1cb2932..8e627dcb7 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/Config.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/Config.kt
@@ -1,9 +1,8 @@
 package com.xebia.functional.xef
 
 import arrow.core.nonEmptyListOf
-import com.xebia.functional.openai.Config as OpenAIConfig
-import com.xebia.functional.openai.generated.api.OpenAI
 import com.xebia.functional.xef.env.getenv
+import com.xebia.functional.xef.openapi.OpenAI
 import io.ktor.client.*
 import io.ktor.client.engine.*
 import io.ktor.client.plugins.*
@@ -74,15 +73,5 @@ fun OpenAI(
     }
   }
   val client = httpClientEngine?.let { HttpClient(it, clientConfig) } ?: HttpClient(clientConfig)
-  return OpenAI(
-    client,
-    OpenAIConfig(
-      baseUrl = config.baseUrl,
-      token = token,
-      org = config.org,
-      json = config.json,
-      streamingPrefix = config.streamingPrefix,
-      streamingDelimiter = config.streamingDelimiter
-    )
-  )
+  return OpenAI(client)
 }
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/Tool.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/Tool.kt
index a0b20a62e..0db21d15b 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/Tool.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/Tool.kt
@@ -1,10 +1,10 @@
 package com.xebia.functional.xef
 
-import com.xebia.functional.openai.generated.model.FunctionObject
 import com.xebia.functional.xef.conversation.Description
 import com.xebia.functional.xef.llm.FunctionCall
 import com.xebia.functional.xef.llm.StreamedFunction
 import com.xebia.functional.xef.llm.chatFunction
+import com.xebia.functional.xef.openapi.FunctionObject
 import kotlin.jvm.JvmName
 import kotlin.reflect.KClass
 import kotlin.reflect.KFunction1
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Chat.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Chat.kt
index f8a39155f..e1cca7998 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Chat.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Chat.kt
@@ -1,9 +1,5 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequest
-import com.xebia.functional.openai.generated.model.CreateChatCompletionResponse
-import com.xebia.functional.openai.generated.model.CreateChatCompletionResponseChoicesInner
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.Tool
 import com.xebia.functional.xef.conversation.AiDsl
@@ -11,6 +7,9 @@ import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.llm.models.MessageWithUsage
 import com.xebia.functional.xef.llm.models.MessagesUsage
 import com.xebia.functional.xef.llm.models.MessagesWithUsage
+import com.xebia.functional.xef.openapi.Chat
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.CreateChatCompletionResponse
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder
 import com.xebia.functional.xef.store.Memory
@@ -38,7 +37,8 @@ fun Chat.promptStreaming(
 
   val buffer = StringBuilder()
 
-  this@promptStreaming.createChatCompletionStream(request)
+  completions
+    .createChatCompletionStream(request)
     .mapNotNull {
       val content = it.choices.firstOrNull()?.delta?.content
       if (content != null) {
@@ -88,7 +88,7 @@ suspend fun Chat.promptMessagesAndUsage(
 private suspend fun <T> Chat.promptResponse(
   prompt: Prompt,
   scope: Conversation = Conversation(),
-  block: suspend Chat.(CreateChatCompletionResponseChoicesInner) -> T?
+  block: suspend Chat.(CreateChatCompletionResponse.Choices) -> T?
 ): Pair<List<T>, CreateChatCompletionResponse> =
   scope.metric.promptSpan(prompt) {
     val promptMemories: List<Memory> = prompt.messages.toMemory(scope)
@@ -107,7 +107,7 @@ private suspend fun <T> Chat.promptResponse(
         seed = adaptedPrompt.configuration.seed,
       )
 
-    val createResponse: CreateChatCompletionResponse = createChatCompletion(request)
+    val createResponse = completions.createChatCompletion(request)
     Pair(
       createResponse
         .addMetrics(scope)
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/ChatWithFunctions.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/ChatWithFunctions.kt
index 269b23da3..7cca85a0c 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/ChatWithFunctions.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/ChatWithFunctions.kt
@@ -3,14 +3,17 @@ package com.xebia.functional.xef.llm
 import arrow.core.nonFatalOrThrow
 import arrow.core.raise.catch
 import arrow.fx.coroutines.parMapNotNull
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.AIEvent
 import com.xebia.functional.xef.Tool
 import com.xebia.functional.xef.conversation.AiDsl
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.llm.models.functions.buildJsonSchema
+import com.xebia.functional.xef.openapi.*
+import com.xebia.functional.xef.openapi.Chat
+import com.xebia.functional.xef.openapi.ChatCompletionToolChoiceOption.NoneOrAutoOrRequired.Auto
+import com.xebia.functional.xef.openapi.ChatCompletionToolChoiceOption.NoneOrAutoOrRequired.Required
+import com.xebia.functional.xef.openapi.FunctionObject
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.tool
 import io.github.oshai.kotlinlogging.KotlinLogging
@@ -155,17 +158,17 @@ private fun assistantRequestedCallMessage(
 ): ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage =
   ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage(
     ChatCompletionRequestAssistantMessage(
-      role = ChatCompletionRequestAssistantMessage.Role.assistant,
+      role = ChatCompletionRequestAssistantMessage.Role.Assistant,
       toolCalls =
         calls.map {
           ChatCompletionMessageToolCall(
             id = it.callId,
             function =
-              ChatCompletionMessageToolCallFunction(
+              ChatCompletionMessageToolCall.Function(
                 name = it.functionName,
                 arguments = it.arguments
               ),
-            type = ChatCompletionMessageToolCall.Type.function
+            type = ChatCompletionMessageToolCall.Type.Function
           )
         }
     )
@@ -229,7 +232,7 @@ private suspend fun <A> Chat.promptWithResponse(
     val request = createChatCompletionRequest(adaptedPrompt)
     tryDeserialize(serializer, promptWithFunctions.configuration.maxDeserializationAttempts) {
       val requestedMemories = prompt.messages.toMemory(scope)
-      val response = createChatCompletion(request).addMetrics(scope)
+      val response = completions.createChatCompletion(request).addMetrics(scope)
       response.choices.addChoiceWithFunctionsToMemory(
         scope,
         requestedMemories,
@@ -256,20 +259,20 @@ private fun chatCompletionToolChoiceOption(adaptedPrompt: Prompt): ChatCompletio
   if (adaptedPrompt.functions.size == 1)
     ChatCompletionToolChoiceOption.CaseChatCompletionNamedToolChoice(
       ChatCompletionNamedToolChoice(
-        type = ChatCompletionNamedToolChoice.Type.function,
+        type = ChatCompletionNamedToolChoice.Type.Function,
         // TODO review access to first
-        function = ChatCompletionNamedToolChoiceFunction(adaptedPrompt.functions.first().name)
+        function = ChatCompletionNamedToolChoice.Function(adaptedPrompt.functions.first().name)
       )
     )
   else {
-    if (adaptedPrompt.model is CreateChatCompletionRequestModel.Custom)
-      ChatCompletionToolChoiceOption.CaseString("auto")
-    else ChatCompletionToolChoiceOption.CaseString("required")
+    if (adaptedPrompt.model is CreateChatCompletionRequest.Model.OpenCase)
+      ChatCompletionToolChoiceOption.CaseNoneOrAutoOrRequired(Auto)
+    else ChatCompletionToolChoiceOption.CaseNoneOrAutoOrRequired(Required)
   }
 
 private fun chatCompletionTools(adaptedPrompt: Prompt): List<ChatCompletionTool> =
   adaptedPrompt.functions.map {
-    ChatCompletionTool(type = ChatCompletionTool.Type.function, function = it)
+    ChatCompletionTool(type = ChatCompletionTool.Type.Function, function = it)
   }
 
 @AiDsl
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Embeddings.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Embeddings.kt
index 981d52fa5..d68810e6c 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Embeddings.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Embeddings.kt
@@ -1,17 +1,13 @@
 package com.xebia.functional.xef.llm
 
 import arrow.fx.coroutines.parMap
-import com.xebia.functional.openai.generated.api.Embeddings
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequest
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestInput
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
+import com.xebia.functional.xef.openapi.*
 
 suspend fun Embeddings.embedDocuments(
   texts: List<String>,
   chunkSize: Int = 400,
-  embeddingRequestModel: CreateEmbeddingRequestModel =
-    CreateEmbeddingRequestModel.text_embedding_ada_002
+  embeddingRequestModel: CreateEmbeddingRequest.Model =
+    CreateEmbeddingRequest.Model.TextEmbeddingAda002
 ): List<Embedding> =
   if (texts.isEmpty()) emptyList()
   else
@@ -21,7 +17,7 @@ suspend fun Embeddings.embedDocuments(
         createEmbedding(
             CreateEmbeddingRequest(
               model = embeddingRequestModel,
-              input = CreateEmbeddingRequestInput.CaseStrings(it)
+              input = CreateEmbeddingRequest.Input.CaseStrings(it)
             )
           )
           .data
@@ -30,7 +26,7 @@ suspend fun Embeddings.embedDocuments(
 
 suspend fun Embeddings.embedQuery(
   text: String,
-  embeddingRequestModel: CreateEmbeddingRequestModel
+  embeddingRequestModel: CreateEmbeddingRequest.Model
 ): List<Embedding> =
   if (text.isNotEmpty())
     embedDocuments(texts = listOf(text), embeddingRequestModel = embeddingRequestModel)
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/LLM.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/LLM.kt
index 928de8425..fe79bf5c8 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/LLM.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/LLM.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
 import com.xebia.functional.tokenizer.Encoding
 import com.xebia.functional.tokenizer.ModelType
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
 import com.xebia.functional.xef.prompt.completionRole
 import com.xebia.functional.xef.prompt.contentAsString
 
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MemoryManagement.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MemoryManagement.kt
index 9ea8a63ed..df9a65249 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MemoryManagement.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MemoryManagement.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.conversation.MessagesToHistory
+import com.xebia.functional.xef.openapi.*
 import com.xebia.functional.xef.store.ConversationId
 import com.xebia.functional.xef.store.MemorizedMessage
 import com.xebia.functional.xef.store.Memory
@@ -32,11 +32,11 @@ internal fun List<ChatCompletionRequestMessage>.toMemory(scope: Conversation): L
   } else emptyList()
 }
 
-internal suspend fun List<CreateChatCompletionResponseChoicesInner>.addChoiceWithFunctionsToMemory(
+internal suspend fun List<CreateChatCompletionResponse.Choices>.addChoiceWithFunctionsToMemory(
   scope: Conversation,
   previousMemories: List<Memory>,
   history: MessagesToHistory
-): List<CreateChatCompletionResponseChoicesInner> = also {
+): List<CreateChatCompletionResponse.Choices> = also {
   val cid = scope.conversationId
   if (history != MessagesToHistory.NONE && isNotEmpty() && cid != null) {
     val aiMemory = this.map { it.message.toMemory(cid, scope.store.incrementIndexAndGet()) }
@@ -45,11 +45,11 @@ internal suspend fun List<CreateChatCompletionResponseChoicesInner>.addChoiceWit
   }
 }
 
-internal suspend fun List<CreateChatCompletionResponseChoicesInner>.addChoiceToMemory(
+internal suspend fun List<CreateChatCompletionResponse.Choices>.addChoiceToMemory(
   scope: Conversation,
   previousMemories: List<Memory>,
   history: MessagesToHistory
-): List<CreateChatCompletionResponseChoicesInner> = also {
+): List<CreateChatCompletionResponse.Choices> = also {
   val cid = scope.conversationId
   if (history != MessagesToHistory.NONE && isNotEmpty() && cid != null) {
     val aiMemory =
@@ -65,10 +65,10 @@ suspend fun VectorStore.addMemoriesByHistory(history: MessagesToHistory, memorie
       addMemories(memories)
     }
     MessagesToHistory.ONLY_SYSTEM_MESSAGES -> {
-      addMemories(memories.filter { it.content.role == ChatCompletionRole.system })
+      addMemories(memories.filter { it.content.role == ChatCompletionRole.System })
     }
     MessagesToHistory.NOT_SYSTEM_MESSAGES -> {
-      addMemories(memories.filter { it.content.role != ChatCompletionRole.system })
+      addMemories(memories.filter { it.content.role != ChatCompletionRole.System })
     }
     MessagesToHistory.NONE -> {}
   }
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MetricManagement.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MetricManagement.kt
index 91d8e948a..806e42c53 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MetricManagement.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/MetricManagement.kt
@@ -1,12 +1,12 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionResponse
-import com.xebia.functional.openai.generated.model.MessageObject
-import com.xebia.functional.openai.generated.model.RunObject
-import com.xebia.functional.openai.generated.model.RunStepObject
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.llm.assistants.RunDelta
 import com.xebia.functional.xef.metrics.Metric
+import com.xebia.functional.xef.openapi.CreateChatCompletionResponse
+import com.xebia.functional.xef.openapi.MessageObject
+import com.xebia.functional.xef.openapi.RunObject
+import com.xebia.functional.xef.openapi.RunStepObject
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.completionRole
 import com.xebia.functional.xef.prompt.contentAsString
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/PromptCalculator.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/PromptCalculator.kt
index f8053056f..02d81d490 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/PromptCalculator.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/PromptCalculator.kt
@@ -1,10 +1,10 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.conversation.MessagesFromHistory
 import com.xebia.functional.xef.llm.models.modelType
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.store.Memory
 
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/StreamedFunction.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/StreamedFunction.kt
index 6164b8a1d..e9d454957 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/StreamedFunction.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/StreamedFunction.kt
@@ -1,9 +1,8 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.llm.StreamedFunction.Companion.PropertyType.*
+import com.xebia.functional.xef.openapi.*
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder
 import kotlin.jvm.JvmSynthetic
@@ -48,7 +47,7 @@ sealed class StreamedFunction<out A> {
     ) {
       val messages = mutableListOf<ChatCompletionRequestMessage>()
       // this function call is mutable and will be updated as the stream progresses
-      var functionCall = ChatCompletionMessageToolCallFunction("", "")
+      var functionCall = ChatCompletionMessageToolCall.Function("", "")
       // the current property is mutable and will be updated as the stream progresses
       var currentProperty: String? = null
       // we keep track to not emit the same property multiple times
@@ -61,7 +60,7 @@ sealed class StreamedFunction<out A> {
       // as the LLM is sending us chunks with malformed JSON
       if (schema != null) {
         val example = createExampleFromSchema(schema)
-        chat
+        chat.completions
           .createChatCompletionStream(request)
           .onCompletion {
             val newMessages = prompt.messages + messages
@@ -128,7 +127,7 @@ sealed class StreamedFunction<out A> {
     }
 
     private suspend fun <A> FlowCollector<StreamedFunction<A>>.streamResult(
-      functionCall: ChatCompletionMessageToolCallFunction,
+      functionCall: ChatCompletionMessageToolCall.Function,
       messages: MutableList<ChatCompletionRequestMessage>,
       serializer: suspend (json: String) -> A
     ) {
@@ -290,13 +289,13 @@ sealed class StreamedFunction<out A> {
     }
 
     private fun mergeArgumentsWithDelta(
-      functionCall: ChatCompletionMessageToolCallFunction,
+      functionCall: ChatCompletionMessageToolCall.Function,
       functionCall0: ChatCompletionMessageToolCallChunk
-    ): ChatCompletionMessageToolCallFunction =
+    ): ChatCompletionMessageToolCall.Function =
       functionCall.copy(arguments = functionCall.arguments + (functionCall0.function?.arguments))
 
     private fun getLastReferencedPropertyInArguments(
-      functionCall: ChatCompletionMessageToolCallFunction
+      functionCall: ChatCompletionMessageToolCall.Function
     ): String? =
       """"(.*?)":"""
         .toRegex()
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Vision.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Vision.kt
index 38d878838..efe94eede 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Vision.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/Vision.kt
@@ -1,11 +1,8 @@
 package com.xebia.functional.xef.llm
 
-import com.xebia.functional.openai.UploadFile
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.Image
 import com.xebia.functional.xef.Tool
 import com.xebia.functional.xef.conversation.Conversation
+import com.xebia.functional.xef.openapi.*
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.image
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.user
@@ -21,9 +18,9 @@ suspend inline fun <reified A> Chat.visionStructured(
   prompt: String,
   url: String,
   conversation: Conversation = Conversation(),
-  model: CreateChatCompletionRequestModel = CreateChatCompletionRequestModel.gpt_4_vision_preview,
-  functionsModel: CreateChatCompletionRequestModel =
-    CreateChatCompletionRequestModel.gpt_3_5_turbo_0125,
+  model: CreateChatCompletionRequest.Model = CreateChatCompletionRequest.Model.Gpt4VisionPreview,
+  functionsModel: CreateChatCompletionRequest.Model =
+    CreateChatCompletionRequest.Model.Gpt35Turbo0125,
   tools: List<Tool<*>> = emptyList()
 ): A {
   val response = vision(prompt, url, model, conversation).toList().joinToString("") { it }
@@ -38,7 +35,7 @@ suspend inline fun <reified A> Chat.visionStructured(
 fun Chat.vision(
   prompt: String,
   url: String,
-  model: CreateChatCompletionRequestModel = CreateChatCompletionRequestModel.gpt_4_vision_preview,
+  model: CreateChatCompletionRequest.Model = CreateChatCompletionRequest.Model.Gpt4VisionPreview,
   conversation: Conversation = Conversation()
 ): Flow<String> =
   promptStreaming(
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Assistant.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Assistant.kt
index 5d84096d4..994e67d91 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Assistant.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Assistant.kt
@@ -1,11 +1,11 @@
 package com.xebia.functional.xef.llm.assistants
 
-import com.xebia.functional.openai.generated.api.Assistants
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.Config
 import com.xebia.functional.xef.OpenAI
 import com.xebia.functional.xef.llm.assistants.AssistantThread.Companion.defaultConfig
 import com.xebia.functional.xef.llm.models.functions.buildJsonSchema
+import com.xebia.functional.xef.openapi.*
+import com.xebia.functional.xef.openapi.Assistants
 import io.ktor.util.logging.*
 import kotlinx.serialization.KSerializer
 import kotlinx.serialization.Serializable
@@ -74,12 +74,12 @@ class Assistant(
     @Serializable data class ToolOutput(val schema: JsonObject, val result: JsonElement)
 
     suspend operator fun invoke(
-      model: CreateAssistantRequestModel,
+      model: CreateAssistantRequest.Model,
       name: String? = null,
       description: String? = null,
       instructions: String? = null,
-      tools: List<AssistantObjectToolsInner> = arrayListOf(),
-      toolResources: CreateAssistantRequestToolResources? = null,
+      tools: List<CreateAssistantRequest.Tools> = arrayListOf(),
+      toolResources: CreateAssistantRequest.ToolResources? = null,
       metadata: JsonObject? = null,
       toolsConfig: List<Tool.Companion.ToolConfig<*, *>> = emptyList(),
       config: Config = Config(),
@@ -120,19 +120,19 @@ class Assistant(
       val fileIds = parsed["file_ids"]?.let { (it as List<*>).map { it.toString() } }
       val vectorStoreIds = parsed["vector_store_ids"]?.let { (it as List<*>).map { it.toString() } }
       val toolResourcesRequest =
-        CreateAssistantRequestToolResources(
+        CreateAssistantRequest.ToolResources(
           codeInterpreter =
-            fileIds?.let { CreateAssistantRequestToolResourcesCodeInterpreter(fileIds = it) },
+            fileIds?.let { CreateAssistantRequest.ToolResources.CodeInterpreter(fileIds = it) },
           fileSearch =
             vectorStoreIds?.let {
-              CreateAssistantRequestToolResourcesFileSearch(vectorStoreIds = it)
+              CreateAssistantRequest.ToolResources.FileSearch(vectorStoreIds = it)
             }
         )
       val assistantRequest =
         AssistantRequest(
           assistantId = parsed["assistant_id"]?.literalContentOrNull,
           model =
-            parsed["model"]?.literalContentOrNull?.let { CreateAssistantRequestModel.Custom(it) }
+            parsed["model"]?.literalContentOrNull?.let { CreateAssistantRequest.Model.OpenCase(it) }
               ?: error("model is required"),
           name = parsed["name"]?.literalContentOrNull,
           description = parsed["description"]?.literalContentOrNull,
@@ -163,7 +163,7 @@ class Assistant(
                             functionObject.name,
                             functionObject.description ?: "",
                             functionObject.parameters?.let { el ->
-                              Json.encodeToString(JsonObject.serializer(), el)
+                              Json.encodeToString(JsonElement.serializer(), el)
                             } ?: ""
                           )
                         } else {
@@ -194,16 +194,22 @@ class Assistant(
             name = assistantRequest.name,
             description = assistantRequest.description,
             instructions = assistantRequest.instructions,
-            tools = assistantTools(assistantRequest),
+            tools =
+              assistantTools(
+                assistantRequest,
+                code = ModifyAssistantRequest.Tools::CaseAssistantToolsCode,
+                fileSearch = ModifyAssistantRequest.Tools::CaseAssistantToolsFileSearch,
+                function = ModifyAssistantRequest.Tools::CaseAssistantToolsFunction
+              ),
             toolResources =
               assistantRequest.toolResources?.let {
-                ModifyAssistantRequestToolResources(
+                ModifyAssistantRequest.ToolResources(
                   codeInterpreter =
                     it.codeInterpreter?.let {
-                      ModifyAssistantRequestToolResourcesCodeInterpreter(fileIds = it.fileIds)
+                      ModifyAssistantRequest.ToolResources.CodeInterpreter(fileIds = it.fileIds)
                     },
                   fileSearch =
-                    ModifyAssistantRequestToolResourcesFileSearch(
+                    ModifyAssistantRequest.ToolResources.FileSearch(
                       vectorStoreIds = it.fileSearch?.vectorStoreIds
                     )
                 )
@@ -219,7 +225,13 @@ class Assistant(
               name = assistantRequest.name,
               description = assistantRequest.description,
               instructions = assistantRequest.instructions,
-              tools = assistantTools(assistantRequest),
+              tools =
+                assistantTools(
+                  assistantRequest,
+                  code = CreateAssistantRequest.Tools::CaseAssistantToolsCode,
+                  fileSearch = CreateAssistantRequest.Tools::CaseAssistantToolsFileSearch,
+                  function = CreateAssistantRequest.Tools::CaseAssistantToolsFunction
+                ),
               toolResources = assistantRequest.toolResources,
               metadata =
                 assistantRequest.metadata
@@ -232,23 +244,24 @@ class Assistant(
         )
     }
 
-    private fun assistantTools(
-      assistantRequest: AssistantRequest
-    ): List<AssistantObjectToolsInner> =
+    private fun <A> assistantTools(
+      assistantRequest: AssistantRequest,
+      code: (AssistantToolsCode) -> A,
+      fileSearch: (AssistantToolsFileSearch) -> A,
+      function: (AssistantToolsFunction) -> A
+    ): List<A> =
       assistantRequest.tools.orEmpty().map {
         when (it) {
           is AssistantTool.CodeInterpreter ->
-            AssistantObjectToolsInner.CaseAssistantToolsCode(
-              AssistantToolsCode(type = AssistantToolsCode.Type.code_interpreter)
-            )
+            code(AssistantToolsCode(type = AssistantToolsCode.Type.CodeInterpreter))
+
+          // TODO Retrieval seems to be renamed to file search??
           is AssistantTool.Retrieval ->
-            AssistantObjectToolsInner.CaseAssistantToolsFileSearch(
-              AssistantToolsFileSearch(type = AssistantToolsFileSearch.Type.file_search)
-            )
+            fileSearch(AssistantToolsFileSearch(type = AssistantToolsFileSearch.Type.FileSearch))
           is AssistantTool.Function ->
-            AssistantObjectToolsInner.CaseAssistantToolsFunction(
+            function(
               AssistantToolsFunction(
-                type = AssistantToolsFunction.Type.function,
+                type = AssistantToolsFunction.Type.Function,
                 function =
                   FunctionObject(
                     name = it.name,
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantRequest.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantRequest.kt
index 16d50f0b4..3f9d3dd9f 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantRequest.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantRequest.kt
@@ -1,7 +1,6 @@
 package com.xebia.functional.xef.llm.assistants
 
-import com.xebia.functional.openai.generated.model.CreateAssistantRequestModel
-import com.xebia.functional.openai.generated.model.CreateAssistantRequestToolResources
+import com.xebia.functional.xef.openapi.CreateAssistantRequest
 import kotlinx.serialization.Required
 import kotlinx.serialization.SerialName
 import kotlinx.serialization.Serializable
@@ -9,7 +8,7 @@ import kotlinx.serialization.Serializable
 @Serializable
 data class AssistantRequest(
   @SerialName(value = "assistant_id") val assistantId: String? = null,
-  @SerialName(value = "model") @Required val model: CreateAssistantRequestModel,
+  @SerialName(value = "model") @Required val model: CreateAssistantRequest.Model,
 
   /* The name of the assistant. The maximum length is 256 characters.  */
   @SerialName(value = "name") val name: String? = null,
@@ -23,7 +22,7 @@ data class AssistantRequest(
   /* A list of tool enabled on the assistant. There can be a maximum of 128 tools per assistant. Tools can be of types `code_interpreter`, `retrieval`, or `function`.  */
   @SerialName(value = "tools") val tools: List<AssistantTool>? = arrayListOf(),
   @SerialName(value = "tool_resources")
-  val toolResources: CreateAssistantRequestToolResources? = null,
+  val toolResources: CreateAssistantRequest.ToolResources? = null,
 
   /* Set of 16 key-value pairs that can be attached to an object. This can be useful for storing additional information about the object in a structured format. Keys can be a maximum of 64 characters long and values can be a maxium of 512 characters long.  */
   @SerialName(value = "metadata") val metadata: Map<String, String>? = null
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantThread.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantThread.kt
index 934a1b790..dae87a9d6 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantThread.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/AssistantThread.kt
@@ -1,12 +1,12 @@
 package com.xebia.functional.xef.llm.assistants
 
 import arrow.fx.coroutines.parMapNotNull
-import com.xebia.functional.openai.generated.api.Assistants
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.Config
 import com.xebia.functional.xef.OpenAI
 import com.xebia.functional.xef.llm.addMetrics
 import com.xebia.functional.xef.metrics.Metric
+import com.xebia.functional.xef.openapi.*
+import com.xebia.functional.xef.openapi.Threads.Messages.ListMessagesOrder
 import io.ktor.client.request.*
 import kotlin.jvm.JvmName
 import kotlinx.coroutines.flow.*
@@ -18,7 +18,7 @@ class AssistantThread(
   val threadId: String,
   val metric: Metric = Metric.EMPTY,
   private val config: Config = Config(),
-  private val api: Assistants = OpenAI(config).assistants
+  private val api: Threads = OpenAI(config).threads
 ) {
 
   suspend fun delete(): Boolean =
@@ -31,29 +31,29 @@ class AssistantThread(
     createMessage(
       request =
         CreateMessageRequest(
-          role = CreateMessageRequest.Role.user,
-          content = CreateMessageRequestContent.CaseString(message.content),
-          attachments = message.fileIds.map { MessageObjectAttachmentsInner(fileId = it) }
+          role = CreateMessageRequest.Role.User,
+          content = CreateMessageRequest.Content.CaseString(message.content),
+          attachments = message.fileIds.map { CreateMessageRequest.Attachments(fileId = it) }
         ),
     )
 
   suspend fun createMessage(content: String): MessageObject =
     createMessage(
       CreateMessageRequest(
-        role = CreateMessageRequest.Role.user,
-        content = CreateMessageRequestContent.CaseString(content)
+        role = CreateMessageRequest.Role.User,
+        content = CreateMessageRequest.Content.CaseString(content)
       )
     )
 
   suspend fun createMessage(request: CreateMessageRequest): MessageObject =
-    api.createMessage(threadId, request, configure = ::defaultConfig)
+    api.messages.createMessage(threadId, request, configure = ::defaultConfig)
 
   suspend fun getMessage(messageId: String): MessageObject =
-    api.getMessage(threadId, messageId, configure = ::defaultConfig)
+    api.messages.getMessage(threadId, messageId, configure = ::defaultConfig)
 
   data class ThreadMessagesFilter(
     val limit: Int? = 20,
-    val order: Assistants.OrderListMessages? = Assistants.OrderListMessages.desc,
+    val order: ListMessagesOrder? = ListMessagesOrder.Desc,
     val after: String? = null,
     val before: String? = null
   )
@@ -61,7 +61,7 @@ class AssistantThread(
   suspend fun listMessages(
     filter: ThreadMessagesFilter = ThreadMessagesFilter()
   ): List<MessageObject> =
-    api
+    api.messages
       .listMessages(
         threadId = threadId,
         limit = filter.limit,
@@ -76,10 +76,12 @@ class AssistantThread(
     createRun(CreateRunRequest(assistantId = assistant.assistantId))
 
   suspend fun createRun(request: CreateRunRequest): RunObject =
-    api.createRun(threadId, request, configure = ::defaultConfig).addMetrics(metric, "RunCreated")
+    api.runs
+      .createRun(threadId, request, configure = ::defaultConfig)
+      .addMetrics(metric, "RunCreated")
 
   fun createRunStream(assistant: Assistant, request: CreateRunRequest): Flow<RunDelta> = flow {
-    api
+    api.runs
       .createRunStream(threadId, request, configure = ::defaultConfig)
       .map { RunDelta.fromServerSentEvent(it) }
       .map { it.addMetrics(metric) }
@@ -105,8 +107,8 @@ class AssistantThread(
     flowCollector: FlowCollector<RunDelta>
   ) {
     if (
-      event.run.status == RunObject.Status.requires_action &&
-        event.run.requiredAction?.type == RunObjectRequiredAction.Type.submit_tool_outputs
+      event.run.status == RunObject.Status.RequiresAction &&
+        event.run.requiredAction?.type == RunObject.RequiredAction.Type.SubmitToolOutputs
     ) {
       val calls = event.run.requiredAction?.submitToolOutputs?.toolCalls.orEmpty()
       val callsResult: List<Pair<String, Assistant.Companion.ToolOutput>> =
@@ -116,18 +118,18 @@ class AssistantThread(
         SubmitToolOutputsRunRequest(
           toolOutputs =
             results.map { (toolCallId, result) ->
-              SubmitToolOutputsRunRequestToolOutputsInner(
+              SubmitToolOutputsRunRequest.ToolOutputs(
                 toolCallId = toolCallId,
                 output = Json.encodeToString(Assistant.Companion.ToolOutput.serializer(), result)
               )
             }
         )
 
-      api
+      api.runs.submitToolOutputs
         .submitToolOuputsToRunStream(
           threadId = threadId,
           runId = event.run.id,
-          submitToolOutputsRunRequest = toolOutputsRequest,
+          body = toolOutputsRequest,
           configure = ::defaultConfig
         )
         .collect {
@@ -144,21 +146,21 @@ class AssistantThread(
       val run = getRun(event.run.id)
       val finalEvent =
         when (run.status) {
-          RunObject.Status.queued -> Pair(RunDelta.RunQueued(run), "RunQueued")
-          RunObject.Status.in_progress -> Pair(RunDelta.RunInProgress(run), "RunInProgress")
-          RunObject.Status.requires_action ->
+          RunObject.Status.Queued -> Pair(RunDelta.RunQueued(run), "RunQueued")
+          RunObject.Status.InProgress -> Pair(RunDelta.RunInProgress(run), "RunInProgress")
+          RunObject.Status.RequiresAction ->
             Pair(RunDelta.RunRequiresAction(run), "RunRequiresAction")
-          RunObject.Status.cancelling -> Pair(RunDelta.RunCancelling(run), "RunCancelling")
-          RunObject.Status.cancelled -> Pair(RunDelta.RunCancelled(run), "RunCancelled")
-          RunObject.Status.failed -> Pair(RunDelta.RunFailed(run), "RunFailed")
-          RunObject.Status.completed -> Pair(RunDelta.RunCompleted(run), "RunCompleted")
-          RunObject.Status.expired -> Pair(RunDelta.RunExpired(run), "RunExpired")
-          RunObject.Status.incomplete -> Pair(RunDelta.RunIncomplete(run), "RunIncomplete")
+          RunObject.Status.Cancelling -> Pair(RunDelta.RunCancelling(run), "RunCancelling")
+          RunObject.Status.Cancelled -> Pair(RunDelta.RunCancelled(run), "RunCancelled")
+          RunObject.Status.Failed -> Pair(RunDelta.RunFailed(run), "RunFailed")
+          RunObject.Status.Completed -> Pair(RunDelta.RunCompleted(run), "RunCompleted")
+          RunObject.Status.Expired -> Pair(RunDelta.RunExpired(run), "RunExpired")
+          RunObject.Status.Incomplete -> Pair(RunDelta.RunIncomplete(run), "RunIncomplete")
         }
       flowCollector.emit(finalEvent.first)
       metric.assistantCreateRun(run, finalEvent.second)
 
-      if (run.status == RunObject.Status.requires_action) {
+      if (run.status == RunObject.Status.RequiresAction) {
         takeRequiredAction(
           depth + 1,
           RunDelta.RunRequiresAction(run),
@@ -178,13 +180,9 @@ class AssistantThread(
       val function = toolCall.function
       val functionName = function.name
       val functionArguments = function.arguments
-      return if (functionName != null && functionArguments != null) {
-        val result = assistant.getToolRegistered(functionName, functionArguments)
-        val callId = toolCall.id
-        if (callId != null) {
-          callId to result
-        } else null
-      } else null
+      val result = assistant.getToolRegistered(functionName, functionArguments)
+      val callId = toolCall.id
+      return callId to result
     } catch (e: Throwable) {
       toolCall.id to
         Assistant.Companion.ToolOutput(
@@ -195,22 +193,21 @@ class AssistantThread(
   }
 
   suspend fun getRun(runId: String): RunObject =
-    api.getRun(threadId, runId, configure = ::defaultConfig)
+    api.runs.getRun(threadId, runId, configure = ::defaultConfig)
 
   fun run(assistant: Assistant): Flow<RunDelta> =
     createRunStream(assistant, CreateRunRequest(assistantId = assistant.assistantId))
 
   suspend fun cancelRun(runId: String): RunObject =
-    api.cancelRun(threadId, runId, configure = ::defaultConfig)
+    api.runs.cancel.cancelRun(threadId, runId, configure = ::defaultConfig)
 
   suspend fun runSteps(runId: String): List<RunStepObject> =
-    api.listRunSteps(threadId, runId, configure = ::defaultConfig).data
+    api.runs.steps.listRunSteps(threadId, runId, configure = ::defaultConfig).data
 
-  private fun RunStepObjectStepDetails.toolCalls():
-    List<RunStepDetailsToolCallsObjectToolCallsInner> =
+  private fun RunStepObject.StepDetails.toolCalls(): List<RunStepDetailsToolCallsObject.ToolCalls> =
     when (val step = this) {
-      is RunStepObjectStepDetails.CaseRunStepDetailsMessageCreationObject -> emptyList()
-      is RunStepObjectStepDetails.CaseRunStepDetailsToolCallsObject -> step.value.toolCalls
+      is RunStepObject.StepDetails.CaseRunStepDetailsMessageCreationObject -> emptyList()
+      is RunStepObject.StepDetails.CaseRunStepDetailsToolCallsObject -> step.value.toolCalls
     }
 
   private suspend fun RunDelta.launchMetricsIfNecessary() {
@@ -268,20 +265,21 @@ class AssistantThread(
       metadata: JsonObject? = null,
       metric: Metric = Metric.EMPTY,
       config: Config = Config(),
-      api: Assistants = OpenAI(config).assistants
+      api: Threads = OpenAI(config).threads
     ): AssistantThread =
       AssistantThread(
         threadId =
           api
             .createThread(
-              createThreadRequest =
+              body =
                 CreateThreadRequest(
                   messages =
                     messages.map {
                       CreateMessageRequest(
-                        role = CreateMessageRequest.Role.user,
-                        content = CreateMessageRequestContent.CaseString(it.content),
-                        attachments = it.fileIds.map { MessageObjectAttachmentsInner(fileId = it) }
+                        role = CreateMessageRequest.Role.User,
+                        content = CreateMessageRequest.Content.CaseString(it.content),
+                        attachments =
+                          it.fileIds.map { CreateMessageRequest.Attachments(fileId = it) }
                       )
                     },
                   metadata = metadata
@@ -300,18 +298,18 @@ class AssistantThread(
       metadata: JsonObject? = null,
       metric: Metric = Metric.EMPTY,
       config: Config = Config(),
-      api: Assistants = OpenAI(config).assistants
+      api: Threads = OpenAI(config).threads
     ): AssistantThread =
       AssistantThread(
         api
           .createThread(
-            createThreadRequest =
+            body =
               CreateThreadRequest(
                 messages =
                   messages.map {
                     CreateMessageRequest(
-                      role = CreateMessageRequest.Role.user,
-                      content = CreateMessageRequestContent.CaseString(it)
+                      role = CreateMessageRequest.Role.User,
+                      content = CreateMessageRequest.Content.CaseString(it)
                     )
                   },
                 metadata = metadata
@@ -330,7 +328,7 @@ class AssistantThread(
       metadata: JsonObject? = null,
       metric: Metric = Metric.EMPTY,
       config: Config = Config(),
-      api: Assistants = OpenAI(config).assistants
+      api: Threads = OpenAI(config).threads
     ): AssistantThread =
       AssistantThread(
         api
@@ -348,7 +346,7 @@ class AssistantThread(
       request: CreateThreadRequest,
       metric: Metric = Metric.EMPTY,
       config: Config = Config(),
-      api: Assistants = OpenAI(config).assistants
+      api: Threads = OpenAI(config).threads
     ): AssistantThread =
       AssistantThread(
         api.createThread(request, configure = ::defaultConfig).id,
@@ -361,10 +359,10 @@ class AssistantThread(
       request: CreateThreadAndRunRequest,
       metric: Metric = Metric.EMPTY,
       config: Config = Config(),
-      api: Assistants = OpenAI(config).assistants
+      api: Threads = OpenAI(config).threads
     ): AssistantThread =
       AssistantThread(
-        api.createThreadAndRun(request, configure = ::defaultConfig).id,
+        api.runs.createThreadAndRun(request, configure = ::defaultConfig).id,
         metric,
         config,
         api
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/RunDelta.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/RunDelta.kt
index fec396e02..f38d369d6 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/RunDelta.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/RunDelta.kt
@@ -1,8 +1,7 @@
 package com.xebia.functional.xef.llm.assistants
 
-import com.xebia.functional.openai.ServerSentEvent
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.Config
+import com.xebia.functional.xef.openapi.*
 import kotlin.jvm.JvmInline
 import kotlinx.serialization.SerialName
 import kotlinx.serialization.Serializable
@@ -75,7 +74,7 @@ sealed interface RunDelta {
 
   @Serializable
   data class RunStepDeltaObjectInner(
-    @SerialName("step_details") val stepDetails: RunStepObjectStepDetails
+    @SerialName("step_details") val stepDetails: RunStepObject.StepDetails
   )
 
   /** [RunDeltaEvent.thread_run_step_completed] */
@@ -237,19 +236,19 @@ sealed interface RunDelta {
       }
       is RunStepCompleted -> {
         when (val details = runStep.stepDetails) {
-          is RunStepObjectStepDetails.CaseRunStepDetailsMessageCreationObject ->
+          is RunStepObject.StepDetails.CaseRunStepDetailsMessageCreationObject ->
             println("Creating msg: ${details.value.messageCreation.messageId}")
-          is RunStepObjectStepDetails.CaseRunStepDetailsToolCallsObject ->
+          is RunStepObject.StepDetails.CaseRunStepDetailsToolCallsObject ->
             println(
               "Tool calls: ${details.value.toolCalls.map { 
               when (it) {
-                is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsCodeObject -> 
+                is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsCodeObject -> 
                   "Code: ${it.value.codeInterpreter.input}"
-                is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsFunctionObject -> {
+                is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsFunctionObject -> {
                   val function = it.value.function
                   "Function: ${function.name}(${function.arguments})"
                 }
-                is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsFileSearchObject -> {
+                is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsFileSearchObject -> {
                   val retrieval = it.value.fileSearch
                   "Retrieval: $retrieval"
                 }
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Tool.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Tool.kt
index 5789961f0..331c20445 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Tool.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/assistants/Tool.kt
@@ -1,7 +1,7 @@
 package com.xebia.functional.xef.llm.assistants
 
-import com.xebia.functional.openai.generated.model.FunctionObject
 import com.xebia.functional.xef.llm.chatFunction
+import com.xebia.functional.xef.openapi.FunctionObject
 import kotlinx.serialization.KSerializer
 import kotlinx.serialization.serializer
 
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/Messages.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/Messages.kt
index 1817d3f16..4d7db2902 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/Messages.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/Messages.kt
@@ -1,6 +1,6 @@
 package com.xebia.functional.xef.llm.models
 
-import com.xebia.functional.openai.generated.model.CompletionUsage
+import com.xebia.functional.xef.openapi.CompletionUsage
 
 data class MessagesWithUsage(val messages: List<String>, val usage: MessagesUsage?)
 
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/OpenAIClient.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/OpenAIClient.kt
index ba9b824f5..41e4114bf 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/OpenAIClient.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/llm/models/OpenAIClient.kt
@@ -1,17 +1,17 @@
 package com.xebia.functional.xef.llm.models
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
 import com.xebia.functional.tokenizer.ModelType
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.CreateEmbeddingRequest
 
-fun CreateChatCompletionRequestModel.modelType(forFunctions: Boolean = false): ModelType {
+fun CreateChatCompletionRequest.Model.modelType(forFunctions: Boolean = false): ModelType {
   val stringValue = value
   val forFunctionsModel = ModelType.functionSpecific.find { forFunctions && it.name == stringValue }
   return forFunctionsModel
     ?: (ModelType.all.find { it.name == stringValue } ?: ModelType.TODO(stringValue))
 }
 
-fun CreateEmbeddingRequestModel.modelType(): ModelType {
+fun CreateEmbeddingRequest.Model.modelType(): ModelType {
   val stringValue = value
   return ModelType.all.find { it.name == stringValue } ?: ModelType.TODO(stringValue)
 }
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/LogsMetric.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/LogsMetric.kt
index c64c4d194..a151b1967 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/LogsMetric.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/LogsMetric.kt
@@ -1,9 +1,9 @@
 package com.xebia.functional.xef.metrics
 
 import arrow.atomic.AtomicInt
-import com.xebia.functional.openai.generated.model.MessageObject
-import com.xebia.functional.openai.generated.model.RunObject
-import com.xebia.functional.openai.generated.model.RunStepObject
+import com.xebia.functional.xef.openapi.MessageObject
+import com.xebia.functional.xef.openapi.RunObject
+import com.xebia.functional.xef.openapi.RunStepObject
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.contentAsString
 import io.github.oshai.kotlinlogging.KotlinLogging
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/Metric.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/Metric.kt
index e7a1f47f0..3587920dd 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/Metric.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/metrics/Metric.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.metrics
 
-import com.xebia.functional.openai.generated.model.MessageObject
-import com.xebia.functional.openai.generated.model.RunObject
-import com.xebia.functional.openai.generated.model.RunStepObject
+import com.xebia.functional.xef.openapi.MessageObject
+import com.xebia.functional.xef.openapi.RunObject
+import com.xebia.functional.xef.openapi.RunStepObject
 import com.xebia.functional.xef.prompt.Prompt
 
 interface Metric {
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PlatformPromptBuilder.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PlatformPromptBuilder.kt
index 0a80019a0..d39d64227 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PlatformPromptBuilder.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PlatformPromptBuilder.kt
@@ -1,12 +1,12 @@
 package com.xebia.functional.xef.prompt
 
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.FunctionObject
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.FunctionObject
 import com.xebia.functional.xef.prompt.configuration.PromptConfiguration
 
 class PlatformPromptBuilder(
-  private val model: CreateChatCompletionRequestModel,
+  private val model: CreateChatCompletionRequest.Model,
   private val functions: List<FunctionObject>,
   private val configuration: PromptConfiguration
 ) : PromptBuilder {
@@ -21,7 +21,7 @@ class PlatformPromptBuilder(
 
   companion object {
     fun create(
-      model: CreateChatCompletionRequestModel,
+      model: CreateChatCompletionRequest.Model,
       functions: List<FunctionObject>,
       configuration: PromptConfiguration
     ): PlatformPromptBuilder = PlatformPromptBuilder(model, functions, configuration)
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/Prompt.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/Prompt.kt
index 62ad2d355..1db9b6921 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/Prompt.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/Prompt.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.prompt
 
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.FunctionObject
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.FunctionObject
 import com.xebia.functional.xef.prompt.configuration.PromptConfiguration
 import kotlin.jvm.JvmOverloads
 import kotlin.jvm.JvmSynthetic
@@ -14,19 +14,19 @@ import kotlin.jvm.JvmSynthetic
 data class Prompt
 @JvmOverloads
 constructor(
-  val model: CreateChatCompletionRequestModel,
+  val model: CreateChatCompletionRequest.Model,
   val messages: List<ChatCompletionRequestMessage>,
   val functions: List<FunctionObject> = emptyList(),
   val configuration: PromptConfiguration = PromptConfiguration.DEFAULTS
 ) {
 
   constructor(
-    model: CreateChatCompletionRequestModel,
+    model: CreateChatCompletionRequest.Model,
     value: String
   ) : this(model, listOf(PromptBuilder.user(value)), emptyList())
 
   constructor(
-    model: CreateChatCompletionRequestModel,
+    model: CreateChatCompletionRequest.Model,
     value: String,
     configuration: PromptConfiguration
   ) : this(model, listOf(PromptBuilder.user(value)), emptyList(), configuration)
@@ -34,7 +34,7 @@ constructor(
   companion object {
     @JvmSynthetic
     operator fun invoke(
-      model: CreateChatCompletionRequestModel,
+      model: CreateChatCompletionRequest.Model,
       functions: List<FunctionObject> = emptyList(),
       configuration: PromptConfiguration = PromptConfiguration.DEFAULTS,
       block: PlatformPromptBuilder.() -> Unit
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PromptBuilder.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PromptBuilder.kt
index 3f90a791d..58869c0c0 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PromptBuilder.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/PromptBuilder.kt
@@ -1,6 +1,7 @@
 package com.xebia.functional.xef.prompt
 
-import com.xebia.functional.openai.generated.model.*
+import com.xebia.functional.xef.openapi.*
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
 import com.xebia.functional.xef.prompt.configuration.PromptConfiguration
 import kotlin.jvm.JvmSynthetic
 
@@ -26,12 +27,11 @@ interface PromptBuilder {
   ): ChatCompletionRequestMessage {
     val content = "${contentAsString()}\n${message.contentAsString()}"
     return when (completionRole()) {
-      ChatCompletionRole.Supported.system -> system(content)
-      ChatCompletionRole.Supported.user -> user(content)
-      ChatCompletionRole.Supported.assistant -> assistant(content)
-      ChatCompletionRole.Supported.tool -> error("Tool role is not supported")
-      ChatCompletionRole.Supported.function -> error("Function role is not supported")
-      is ChatCompletionRole.Custom -> error("Custom roles are not supported")
+      ChatCompletionRole.System -> system(content)
+      ChatCompletionRole.User -> user(content)
+      ChatCompletionRole.Assistant -> assistant(content)
+      ChatCompletionRole.Tool -> error("Tool role is not supported")
+      ChatCompletionRole.Function -> error("Function role is not supported")
     }
   }
 
@@ -79,7 +79,7 @@ interface PromptBuilder {
   companion object {
 
     operator fun invoke(
-      model: CreateChatCompletionRequestModel,
+      model: CreateChatCompletionRequest.Model,
       functions: List<FunctionObject>,
       configuration: PromptConfiguration
     ): PlatformPromptBuilder = PlatformPromptBuilder.create(model, functions, configuration)
@@ -87,7 +87,7 @@ interface PromptBuilder {
     fun assistant(value: String): ChatCompletionRequestMessage =
       ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage(
         ChatCompletionRequestAssistantMessage(
-          role = ChatCompletionRequestAssistantMessage.Role.assistant,
+          role = ChatCompletionRequestAssistantMessage.Role.Assistant,
           content = value
         )
       )
@@ -95,15 +95,15 @@ interface PromptBuilder {
     fun user(value: String): ChatCompletionRequestMessage =
       ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage(
         ChatCompletionRequestUserMessage(
-          role = ChatCompletionRequestUserMessage.Role.user,
-          content = ChatCompletionRequestUserMessageContent.CaseString(value)
+          role = ChatCompletionRequestUserMessage.Role.User,
+          content = ChatCompletionRequestUserMessage.Content.CaseString(value)
         )
       )
 
     fun tool(toolCallId: String, value: String): ChatCompletionRequestMessage =
       ChatCompletionRequestMessage.CaseChatCompletionRequestToolMessage(
         ChatCompletionRequestToolMessage(
-          role = ChatCompletionRequestToolMessage.Role.tool,
+          role = ChatCompletionRequestToolMessage.Role.Tool,
           content = value,
           toolCallId = toolCallId
         )
@@ -112,7 +112,7 @@ interface PromptBuilder {
     fun system(value: String): ChatCompletionRequestMessage =
       ChatCompletionRequestMessage.CaseChatCompletionRequestSystemMessage(
         ChatCompletionRequestSystemMessage(
-          role = ChatCompletionRequestSystemMessage.Role.system,
+          role = ChatCompletionRequestSystemMessage.Role.System,
           content = value
         )
       )
@@ -120,21 +120,21 @@ interface PromptBuilder {
     fun image(url: String, text: String): ChatCompletionRequestMessage =
       ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage(
         ChatCompletionRequestUserMessage(
-          role = ChatCompletionRequestUserMessage.Role.user,
+          role = ChatCompletionRequestUserMessage.Role.User,
           content =
-            ChatCompletionRequestUserMessageContent.CaseChatCompletionRequestMessageContentParts(
+            ChatCompletionRequestUserMessage.Content.CaseChatCompletionRequestMessageContentParts(
               listOf(
                 ChatCompletionRequestMessageContentPart
                   .CaseChatCompletionRequestMessageContentPartImage(
                     ChatCompletionRequestMessageContentPartImage(
-                      type = ChatCompletionRequestMessageContentPartImage.Type.image_url,
-                      imageUrl = ChatCompletionRequestMessageContentPartImageImageUrl(url)
+                      type = ChatCompletionRequestMessageContentPartImage.Type.ImageUrl,
+                      imageUrl = ChatCompletionRequestMessageContentPartImage.ImageUrl(url)
                     )
                   ),
                 ChatCompletionRequestMessageContentPart
                   .CaseChatCompletionRequestMessageContentPartText(
                     ChatCompletionRequestMessageContentPartText(
-                      type = ChatCompletionRequestMessageContentPartText.Type.text,
+                      type = ChatCompletionRequestMessageContentPartText.Type.Text,
                       text = text
                     )
                   )
@@ -149,8 +149,8 @@ fun ChatCompletionRequestMessage.contentAsString(): String =
   when (this) {
     is ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage ->
       when (val content = value.content) {
-        is ChatCompletionRequestUserMessageContent.CaseString -> content.value
-        is ChatCompletionRequestUserMessageContent.CaseChatCompletionRequestMessageContentParts ->
+        is ChatCompletionRequestUserMessage.Content.CaseString -> content.value
+        is ChatCompletionRequestUserMessage.Content.CaseChatCompletionRequestMessageContentParts ->
           content.value.joinToString {
             when (it) {
               is ChatCompletionRequestMessageContentPart.CaseChatCompletionRequestMessageContentPartImage ->
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/configuration/PromptConfiguration.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/configuration/PromptConfiguration.kt
index 06d8b204c..2237b7fc3 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/configuration/PromptConfiguration.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/prompt/configuration/PromptConfiguration.kt
@@ -1,7 +1,7 @@
 package com.xebia.functional.xef.prompt.configuration
 
-import com.xebia.functional.openai.generated.model.ChatCompletionRole
 import com.xebia.functional.xef.conversation.MessagePolicy
+import com.xebia.functional.xef.openapi.ChatCompletionRole
 import kotlin.jvm.JvmField
 import kotlin.jvm.JvmOverloads
 import kotlinx.serialization.Serializable
@@ -13,7 +13,7 @@ constructor(
   var maxDeserializationAttempts: Int = 3,
   var maxToolCallsPerRound: Int = 10000,
   var concurrentToolCallsPerRound: Int = 5,
-  var user: String = ChatCompletionRole.user.value,
+  var user: String = ChatCompletionRole.User.value,
   var temperature: Double = 0.4,
   var numberOfPredictions: Int = 1,
   var maxTokens: Int = 500,
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/CombinedVectorStore.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/CombinedVectorStore.kt
index 4e10fb1de..97fb955e4 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/CombinedVectorStore.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/CombinedVectorStore.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.store
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
 import com.xebia.functional.xef.llm.models.modelType
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.Embedding
 
 /**
  * A way of composing two [VectorStore] instances together, this class will **first search** [top],
@@ -14,7 +14,7 @@ class CombinedVectorStore(private val top: VectorStore, private val bottom: Vect
   VectorStore by top {
 
   override suspend fun memories(
-    model: CreateChatCompletionRequestModel,
+    model: CreateChatCompletionRequest.Model,
     conversationId: ConversationId,
     limitTokens: Int
   ): List<Memory> {
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/LocalVectorStore.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/LocalVectorStore.kt
index 67908be24..7e8755486 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/LocalVectorStore.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/LocalVectorStore.kt
@@ -4,13 +4,10 @@ import arrow.atomic.Atomic
 import arrow.atomic.AtomicInt
 import arrow.atomic.getAndUpdate
 import arrow.atomic.update
-import com.xebia.functional.openai.generated.api.Embeddings
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
 import com.xebia.functional.xef.llm.embedDocuments
 import com.xebia.functional.xef.llm.embedQuery
 import com.xebia.functional.xef.llm.models.modelType
+import com.xebia.functional.xef.openapi.*
 import kotlin.math.sqrt
 
 private data class State(
@@ -29,12 +26,12 @@ class LocalVectorStore
 private constructor(
   private val embeddings: Embeddings,
   private val state: AtomicState,
-  private val embeddingRequestModel: CreateEmbeddingRequestModel
+  private val embeddingRequestModel: CreateEmbeddingRequest.Model
 ) : VectorStore {
   constructor(
     embeddings: Embeddings,
-    embeddingRequestModel: CreateEmbeddingRequestModel =
-      CreateEmbeddingRequestModel.text_embedding_ada_002
+    embeddingRequestModel: CreateEmbeddingRequest.Model =
+      CreateEmbeddingRequest.Model.TextEmbeddingAda002
   ) : this(embeddings, Atomic(State.empty()), embeddingRequestModel)
 
   override val indexValue: AtomicInt = AtomicInt(0)
@@ -63,7 +60,7 @@ private constructor(
   }
 
   override suspend fun memories(
-    model: CreateChatCompletionRequestModel,
+    model: CreateChatCompletionRequest.Model,
     conversationId: ConversationId,
     limitTokens: Int
   ): List<Memory> {
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/MemorizedMessage.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/MemorizedMessage.kt
index 723690b45..aaf1ea1a4 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/MemorizedMessage.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/MemorizedMessage.kt
@@ -1,13 +1,13 @@
 package com.xebia.functional.xef.store
 
-import com.xebia.functional.openai.generated.model.*
+import com.xebia.functional.xef.openapi.*
 import com.xebia.functional.xef.prompt.completionRole
 
 sealed class MemorizedMessage {
   val role: ChatCompletionRole
     get() =
       when (this) {
-        is Request -> ChatCompletionRole.valueOf(message.completionRole().value)
+        is Request -> message.completionRole()
         is Response -> ChatCompletionRole.valueOf(message.role.name)
       }
 
@@ -17,7 +17,7 @@ sealed class MemorizedMessage {
       is Response ->
         ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage(
           ChatCompletionRequestAssistantMessage(
-            role = ChatCompletionRequestAssistantMessage.Role.assistant,
+            role = ChatCompletionRequestAssistantMessage.Role.Assistant,
             // TODO: Find a new strategy to save the tool calls as content
             content = message.content ?: message.toolCalls?.firstOrNull()?.toString()
           )
@@ -31,32 +31,31 @@ sealed class MemorizedMessage {
 
 fun memorizedMessage(role: ChatCompletionRole, content: String): MemorizedMessage =
   when (role) {
-    ChatCompletionRole.Supported.system ->
+    ChatCompletionRole.System ->
       MemorizedMessage.Request(
         ChatCompletionRequestMessage.CaseChatCompletionRequestSystemMessage(
           ChatCompletionRequestSystemMessage(
             content = content,
-            role = ChatCompletionRequestSystemMessage.Role.system
+            role = ChatCompletionRequestSystemMessage.Role.System
           )
         )
       )
-    ChatCompletionRole.Supported.user ->
+    ChatCompletionRole.User ->
       MemorizedMessage.Request(
         ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage(
           ChatCompletionRequestUserMessage(
-            content = ChatCompletionRequestUserMessageContent.CaseString(content),
-            role = ChatCompletionRequestUserMessage.Role.user
+            content = ChatCompletionRequestUserMessage.Content.CaseString(content),
+            role = ChatCompletionRequestUserMessage.Role.User
           )
         )
       )
-    ChatCompletionRole.Supported.assistant ->
+    ChatCompletionRole.Assistant ->
       MemorizedMessage.Response(
         ChatCompletionResponseMessage(
           content = content,
-          role = ChatCompletionResponseMessage.Role.assistant
+          role = ChatCompletionResponseMessage.Role.Assistant
         )
       )
-    ChatCompletionRole.Supported.tool -> error("Tool messages are not supported")
-    ChatCompletionRole.Supported.function -> error("Function messages are not supported")
-    is ChatCompletionRole.Custom -> error("Custom messages are not supported")
+    ChatCompletionRole.Tool -> error("Tool messages are not supported")
+    ChatCompletionRole.Function -> error("Function messages are not supported")
   }
diff --git a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/VectorStore.kt b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/VectorStore.kt
index 65d13dce8..fb81a9508 100644
--- a/core/src/commonMain/kotlin/com/xebia/functional/xef/store/VectorStore.kt
+++ b/core/src/commonMain/kotlin/com/xebia/functional/xef/store/VectorStore.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.store
 
 import arrow.atomic.AtomicInt
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.Embedding
 import kotlin.jvm.JvmStatic
 
 interface VectorStore {
@@ -16,7 +16,7 @@ interface VectorStore {
   suspend fun addMemories(memories: List<Memory>)
 
   suspend fun memories(
-    model: CreateChatCompletionRequestModel,
+    model: CreateChatCompletionRequest.Model,
     conversationId: ConversationId,
     limitTokens: Int
   ): List<Memory>
@@ -60,7 +60,7 @@ interface VectorStore {
         override suspend fun addMemories(memories: List<Memory>) {}
 
         override suspend fun memories(
-          model: CreateChatCompletionRequestModel,
+          model: CreateChatCompletionRequest.Model,
           conversationId: ConversationId,
           limitTokens: Int
         ): List<Memory> = emptyList()
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/conversation/ConversationSpec.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/conversation/ConversationSpec.kt
index e659386df..b49e1c745 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/conversation/ConversationSpec.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/conversation/ConversationSpec.kt
@@ -1,6 +1,5 @@
 package com.xebia.functional.xef.conversation
 
-import com.xebia.functional.openai.generated.model.*
 import com.xebia.functional.xef.data.TestChatApi
 import com.xebia.functional.xef.data.TestEmbeddings
 import com.xebia.functional.xef.llm.models.modelType
@@ -8,6 +7,7 @@ import com.xebia.functional.xef.llm.promptMessage
 import com.xebia.functional.xef.llm.promptMessages
 import com.xebia.functional.xef.llm.tokensFromMessages
 import com.xebia.functional.xef.metrics.LogsMetric
+import com.xebia.functional.xef.openapi.*
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.assistant
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.system
@@ -28,7 +28,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_4
+      val model = CreateChatCompletionRequest.Model.Gpt4
 
       val scope =
         Conversation(
@@ -66,7 +66,7 @@ class ConversationSpec :
       val vectorStore = scope.store
 
       val chatApi = TestChatApi(responses = messages)
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val totalTokens =
         model.modelType().tokensFromMessages(messages.flatMap(::chatCompletionRequestMessages))
@@ -102,7 +102,7 @@ class ConversationSpec :
       val vectorStore = scope.store
 
       val chatApi = TestChatApi(responses = messages)
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo_16k
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo16k
 
       val totalTokens =
         model.modelType().tokensFromMessages(messages.flatMap(::chatCompletionRequestMessages))
@@ -125,7 +125,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val scope =
         Conversation(
@@ -145,7 +145,7 @@ class ConversationSpec :
       val firstResponse = chatApi.promptMessage(prompt = firstPrompt, scope = scope)
       val aiFirstMessage =
         ChatCompletionRequestAssistantMessage(
-          role = ChatCompletionRequestAssistantMessage.Role.assistant,
+          role = ChatCompletionRequestAssistantMessage.Role.Assistant,
           content = firstResponse
         )
 
@@ -154,7 +154,7 @@ class ConversationSpec :
       val secondResponse = chatApi.promptMessage(prompt = secondPrompt, scope = scope)
       val aiSecondMessage =
         ChatCompletionRequestAssistantMessage(
-          role = ChatCompletionRequestAssistantMessage.Role.assistant,
+          role = ChatCompletionRequestAssistantMessage.Role.Assistant,
           content = secondResponse
         )
 
@@ -167,7 +167,7 @@ class ConversationSpec :
       val thirdResponse = chatApi.promptMessage(prompt = thirdPrompt, scope = scope)
       val aiThirdMessage =
         ChatCompletionRequestAssistantMessage(
-          role = ChatCompletionRequestAssistantMessage.Role.assistant,
+          role = ChatCompletionRequestAssistantMessage.Role.Assistant,
           content = secondResponse
         )
 
@@ -190,7 +190,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -223,7 +223,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -252,7 +252,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -276,14 +276,14 @@ class ConversationSpec :
 
       println(messagesStored)
 
-      messagesStored.filter { it.content.role == ChatCompletionRole.system } shouldBe messagesStored
+      messagesStored.filter { it.content.role == ChatCompletionRole.System } shouldBe messagesStored
     }
 
     "when using MessagesToHistory.NOT_SYSTEM_MESSAGES policy, the scope's store shouldn't contains system messages" {
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -305,14 +305,14 @@ class ConversationSpec :
 
       val messagesStored = scope.store.memories(model, conversationId, Int.MAX_VALUE)
 
-      messagesStored.filter { it.content.role != ChatCompletionRole.system } shouldBe messagesStored
+      messagesStored.filter { it.content.role != ChatCompletionRole.System } shouldBe messagesStored
     }
 
     "when using MessagesToHistory.NONE policy, the scope's store shouldn't contains messages" {
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -334,7 +334,7 @@ class ConversationSpec :
 
       val messagesStored =
         scope.store.memories(model, conversationId, Int.MAX_VALUE).filter {
-          it.content.role == ChatCompletionRole.system
+          it.content.role == ChatCompletionRole.System
         }
 
       messagesStored.size shouldBe 0
@@ -344,7 +344,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -380,7 +380,7 @@ class ConversationSpec :
       val conversationId = ConversationId(UUID.generateUUID().toString())
 
       val chatApi = TestChatApi()
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo
 
       val vectorStore = LocalVectorStore(TestEmbeddings())
 
@@ -417,13 +417,13 @@ private fun chatCompletionRequestMessages(it: Map.Entry<String, String>) =
   listOf(
     ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage(
       ChatCompletionRequestUserMessage(
-        role = ChatCompletionRequestUserMessage.Role.user,
-        content = ChatCompletionRequestUserMessageContent.CaseString(it.key)
+        role = ChatCompletionRequestUserMessage.Role.User,
+        content = ChatCompletionRequestUserMessage.Content.CaseString(it.key)
       )
     ),
     ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage(
       ChatCompletionRequestAssistantMessage(
-        role = ChatCompletionRequestAssistantMessage.Role.assistant,
+        role = ChatCompletionRequestAssistantMessage.Role.Assistant,
         content = it.value
       )
     ),
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestChatApi.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestChatApi.kt
index 99d675727..af0d01914 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestChatApi.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestChatApi.kt
@@ -1,7 +1,7 @@
 package com.xebia.functional.xef.data
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.*
+import com.xebia.functional.xef.openapi.*
+import com.xebia.functional.xef.openapi.Chat
 import com.xebia.functional.xef.prompt.contentAsString
 import io.ktor.client.request.*
 import kotlinx.coroutines.flow.Flow
@@ -10,55 +10,56 @@ class TestChatApi(private val responses: Map<String, String> = emptyMap()) : Cha
 
   var requests: MutableList<CreateChatCompletionRequest> = mutableListOf()
 
-  override suspend fun createChatCompletion(
-    createChatCompletionRequest: CreateChatCompletionRequest,
-    configure: HttpRequestBuilder.() -> Unit
-  ): CreateChatCompletionResponse {
-    requests.add(createChatCompletionRequest)
-    val response =
-      CreateChatCompletionResponse(
-        id = "fake-id",
-        `object` = CreateChatCompletionResponse.Object.chat_completion,
-        created = 0,
-        model = "fake-model",
-        choices =
-          listOf(
-            CreateChatCompletionResponseChoicesInner(
-              message =
-                ChatCompletionResponseMessage(
-                  role = ChatCompletionResponseMessage.Role.assistant,
-                  content =
-                    responses[createChatCompletionRequest.messages.last().contentAsString()]
-                      ?: "fake-content",
-                  toolCalls =
-                    listOf(
-                      ChatCompletionMessageToolCall(
-                        id = "fake-tool-id",
-                        type = ChatCompletionMessageToolCall.Type.function,
-                        function =
-                          ChatCompletionMessageToolCallFunction(
-                            "Answer",
-                            """{ "bar": "fake-answer" }"""
+  override val completions: Chat.Completions =
+    object : Chat.Completions {
+      override suspend fun createChatCompletion(
+        body: CreateChatCompletionRequest,
+        configure: HttpRequestBuilder.() -> Unit
+      ): CreateChatCompletionResponse {
+        requests.add(body)
+        val response =
+          CreateChatCompletionResponse(
+            id = "fake-id",
+            `object` = CreateChatCompletionResponse.Object.ChatCompletion,
+            created = 0,
+            model = "fake-model",
+            choices =
+              listOf(
+                CreateChatCompletionResponse.Choices(
+                  message =
+                    ChatCompletionResponseMessage(
+                      role = ChatCompletionResponseMessage.Role.Assistant,
+                      content = responses[body.messages.last().contentAsString()] ?: "fake-content",
+                      toolCalls =
+                        listOf(
+                          ChatCompletionMessageToolCall(
+                            id = "fake-tool-id",
+                            type = ChatCompletionMessageToolCall.Type.Function,
+                            function =
+                              ChatCompletionMessageToolCall.Function(
+                                "Answer",
+                                """{ "bar": "fake-answer" }"""
+                              )
                           )
-                      )
-                    )
-                ),
-              finishReason = CreateChatCompletionResponseChoicesInner.FinishReason.stop,
-              index = 0,
-              logprobs = null
-            )
-          ),
-        usage = CompletionUsage(0, 0, 0)
-      )
-    return response
-  }
+                        )
+                    ),
+                  finishReason = CreateChatCompletionResponse.Choices.FinishReason.Stop,
+                  index = 0,
+                  logprobs = null
+                )
+              ),
+            usage = CompletionUsage(0, 0, 0)
+          )
+        return response
+      }
 
-  override fun createChatCompletionStream(
-    createChatCompletionRequest: CreateChatCompletionRequest,
-    configure: HttpRequestBuilder.() -> Unit
-  ): Flow<CreateChatCompletionStreamResponse> {
-    throw NotImplementedError("Not implemented")
-  }
+      override suspend fun createChatCompletionStream(
+        body: CreateChatCompletionRequest,
+        configure: HttpRequestBuilder.() -> Unit
+      ): Flow<CreateChatCompletionStreamResponse> {
+        TODO("Not yet implemented")
+      }
+    }
 
   override fun close() {}
 }
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestEmbeddings.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestEmbeddings.kt
index b607b7fc3..5dab226bb 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestEmbeddings.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/data/TestEmbeddings.kt
@@ -1,9 +1,8 @@
 package com.xebia.functional.xef.data
 
-import com.xebia.functional.openai.generated.api.Embeddings
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequest
-import com.xebia.functional.openai.generated.model.CreateEmbeddingResponse
-import com.xebia.functional.openai.generated.model.CreateEmbeddingResponseUsage
+import com.xebia.functional.xef.openapi.CreateEmbeddingRequest
+import com.xebia.functional.xef.openapi.CreateEmbeddingResponse
+import com.xebia.functional.xef.openapi.Embeddings
 import io.ktor.client.request.*
 
 class TestEmbeddings : Embeddings, AutoCloseable {
@@ -19,8 +18,8 @@ class TestEmbeddings : Embeddings, AutoCloseable {
       CreateEmbeddingResponse(
         data = emptyList(),
         model = "",
-        `object` = CreateEmbeddingResponse.Object.list,
-        usage = CreateEmbeddingResponseUsage(0, 0)
+        `object` = CreateEmbeddingResponse.Object.List,
+        usage = CreateEmbeddingResponse.Usage(0, 0)
       )
     return response
   }
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/prompt/PromptBuilderSpec.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/prompt/PromptBuilderSpec.kt
index 45ee78648..1ba9144aa 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/prompt/PromptBuilderSpec.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/prompt/PromptBuilderSpec.kt
@@ -1,6 +1,6 @@
 package com.xebia.functional.xef.prompt
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.assistant
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.system
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.user
@@ -9,7 +9,7 @@ import io.kotest.matchers.shouldBe
 
 class PromptBuilderSpec :
   StringSpec({
-    val model = CreateChatCompletionRequestModel.gpt_4
+    val model = CreateChatCompletionRequest.Model.Gpt4
     "buildPrompt should return the expected messages" {
       val messages =
         Prompt(model) {
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/CombinedVectorStoreSpec.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/CombinedVectorStoreSpec.kt
index 54cca47ab..956cd8b63 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/CombinedVectorStoreSpec.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/CombinedVectorStoreSpec.kt
@@ -1,15 +1,15 @@
 package com.xebia.functional.xef.store
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.data.TestEmbeddings
 import com.xebia.functional.xef.llm.models.modelType
 import com.xebia.functional.xef.llm.tokensFromMessages
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import io.kotest.core.spec.style.StringSpec
 import io.kotest.matchers.shouldBe
 
 class CombinedVectorStoreSpec :
   StringSpec({
-    val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+    val model = CreateChatCompletionRequest.Model.Gpt35Turbo
     "memories function should return all of messages combined in the right order" {
       val memoryData = MemoryData()
 
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/LocalVectorStoreSpec.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/LocalVectorStoreSpec.kt
index abc6a22d4..48afc1ee7 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/LocalVectorStoreSpec.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/LocalVectorStoreSpec.kt
@@ -1,15 +1,15 @@
 package com.xebia.functional.xef.store
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.data.TestEmbeddings
 import com.xebia.functional.xef.llm.models.modelType
 import com.xebia.functional.xef.llm.tokensFromMessages
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import io.kotest.core.spec.style.StringSpec
 import io.kotest.matchers.shouldBe
 
 class LocalVectorStoreSpec :
   StringSpec({
-    val model = CreateChatCompletionRequestModel.gpt_3_5_turbo
+    val model = CreateChatCompletionRequest.Model.Gpt35Turbo
     "memories function should return all of messages in the right order when the limit is greater than the number of stored messages" {
       val localVectorStore = LocalVectorStore(TestEmbeddings())
 
diff --git a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/MemoryData.kt b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/MemoryData.kt
index fd6ed7f02..59b200773 100644
--- a/core/src/commonTest/kotlin/com/xebia/functional/xef/store/MemoryData.kt
+++ b/core/src/commonTest/kotlin/com/xebia/functional/xef/store/MemoryData.kt
@@ -1,10 +1,9 @@
 package com.xebia.functional.xef.store
 
 import arrow.atomic.AtomicInt
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestAssistantMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestUserMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestUserMessageContent
+import com.xebia.functional.xef.openapi.ChatCompletionRequestAssistantMessage
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
+import com.xebia.functional.xef.openapi.ChatCompletionRequestUserMessage
 
 class MemoryData {
   val defaultConversationId = ConversationId("default-id")
@@ -19,15 +18,15 @@ class MemoryData {
     (0 until n).flatMap {
       val m1 =
         ChatCompletionRequestUserMessage(
-          role = ChatCompletionRequestUserMessage.Role.user,
+          role = ChatCompletionRequestUserMessage.Role.User,
           content =
-            ChatCompletionRequestUserMessageContent.CaseString(
+            ChatCompletionRequestUserMessage.Content.CaseString(
               "Question $it${append?.let { ": $it" } ?: ""}"
             )
         )
       val m2 =
         ChatCompletionRequestAssistantMessage(
-          role = ChatCompletionRequestAssistantMessage.Role.assistant,
+          role = ChatCompletionRequestAssistantMessage.Role.Assistant,
           content = "Response $it${append?.let { ": $it" } ?: ""}"
         )
       listOf(
diff --git a/evaluator/build.gradle.kts b/evaluator/build.gradle.kts
index 56ddda484..eb3355939 100644
--- a/evaluator/build.gradle.kts
+++ b/evaluator/build.gradle.kts
@@ -19,7 +19,6 @@ dependencies {
     api(libs.kotlinx.serialization.json)
     detektPlugins(project(":detekt-rules"))
     implementation(projects.xefCore)
-    implementation(projects.xefOpenaiClient)
 }
 
 detekt {
diff --git a/evaluator/src/main/kotlin/com/xebia/functional/xef/evaluator/SuiteBuilder.kt b/evaluator/src/main/kotlin/com/xebia/functional/xef/evaluator/SuiteBuilder.kt
index 832a9d114..26e0755fb 100644
--- a/evaluator/src/main/kotlin/com/xebia/functional/xef/evaluator/SuiteBuilder.kt
+++ b/evaluator/src/main/kotlin/com/xebia/functional/xef/evaluator/SuiteBuilder.kt
@@ -1,10 +1,10 @@
 package com.xebia.functional.xef.evaluator
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.AI
 import com.xebia.functional.xef.AIConfig
 import com.xebia.functional.xef.PromptClassifier
 import com.xebia.functional.xef.evaluator.models.*
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import kotlinx.serialization.SerialName
 import kotlinx.serialization.Serializable
 import kotlinx.serialization.json.Json
@@ -12,7 +12,7 @@ import kotlinx.serialization.serializer
 
 class SuiteBuilder(
   private val description: String,
-  private val model: CreateChatCompletionRequestModel
+  private val model: CreateChatCompletionRequest.Model
 ) {
 
   private val items = mutableListOf<ItemSpec>()
@@ -28,7 +28,7 @@ class SuiteBuilder(
 data class SuiteSpec(
   val description: String,
   val items: List<ItemSpec>,
-  val model: CreateChatCompletionRequestModel
+  val model: CreateChatCompletionRequest.Model
 ) {
 
   suspend inline fun <reified E> evaluate(success: List<E>): SuiteResults<E> where
@@ -62,7 +62,7 @@ data class SuiteSpec(
 
     suspend operator fun invoke(
       description: String,
-      model: CreateChatCompletionRequestModel,
+      model: CreateChatCompletionRequest.Model,
       block: suspend SuiteBuilder.() -> Unit
     ): SuiteSpec = SuiteBuilder(description, model).apply { block() }.build()
 
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/assistants/AssistantStreaming.kt b/examples/src/main/kotlin/com/xebia/functional/xef/assistants/AssistantStreaming.kt
index 8fecf412b..98a4c369e 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/assistants/AssistantStreaming.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/assistants/AssistantStreaming.kt
@@ -1,12 +1,12 @@
 package com.xebia.functional.xef.assistants
 
-import com.xebia.functional.openai.generated.model.CreateRunRequest
 import com.xebia.functional.xef.Config
 import com.xebia.functional.xef.OpenAI
 import com.xebia.functional.xef.llm.assistants.Assistant
 import com.xebia.functional.xef.llm.assistants.AssistantThread
 import com.xebia.functional.xef.llm.assistants.Tool
 import com.xebia.functional.xef.metrics.Metric
+import com.xebia.functional.xef.openapi.CreateRunRequest
 
 suspend fun main() {
 
@@ -19,7 +19,7 @@ suspend fun main() {
       toolsConfig = listOf(Tool.toolOf(SumTool()))
     )
   val config = Config(org = null)
-  val api = OpenAI(config = config, logRequests = false).assistants
+  val api = OpenAI(config = config, logRequests = false).threads
   val thread = AssistantThread(api = api, metric = metric)
   println("Welcome to the Math tutor, ask me anything about math:")
   val userInput = "What is 1+1, explain all the steps and tools you used to solve it."
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/assistants/DSL.kt b/examples/src/main/kotlin/com/xebia/functional/xef/assistants/DSL.kt
index 04c13c081..836188580 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/assistants/DSL.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/assistants/DSL.kt
@@ -5,6 +5,7 @@ import com.xebia.functional.xef.llm.assistants.Assistant
 import com.xebia.functional.xef.llm.assistants.AssistantThread
 import com.xebia.functional.xef.llm.assistants.RunDelta
 import com.xebia.functional.xef.llm.assistants.Tool
+import com.xebia.functional.xef.openapi.OpenAI
 import kotlinx.serialization.Serializable
 
 @Serializable data class SumInput(val left: Int, val right: Int)
@@ -49,7 +50,7 @@ suspend fun main() {
       assistantId = "asst_UxczzpJkysC0l424ood87DAk",
       toolsConfig = listOf(Tool.toolOf(SumTool()))
     )
-  val thread = AssistantThread(api = OpenAI(logRequests = false).assistants, metric = metric)
+  val thread = AssistantThread(api = OpenAI(logRequests = false).threads, metric = metric)
   println("Welcome to the Math tutor, ask me anything about math:")
   while (true) {
     println()
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/conversation/conversations/Animal.kt b/examples/src/main/kotlin/com/xebia/functional/xef/conversation/conversations/Animal.kt
index a43f15e62..02516286b 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/conversation/conversations/Animal.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/conversation/conversations/Animal.kt
@@ -1,6 +1,5 @@
 package com.xebia.functional.xef.conversation.conversations
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.AI
 import com.xebia.functional.xef.AIConfig
 import com.xebia.functional.xef.OpenAI
@@ -8,6 +7,8 @@ import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.conversation.MessagesFromHistory
 import com.xebia.functional.xef.conversation.MessagesToHistory
 import com.xebia.functional.xef.llm.promptMessage
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.OpenAI
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.system
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.user
@@ -36,7 +37,7 @@ suspend fun main() {
       val configNoneFromConversation = PromptConfiguration {
         messagePolicy { addMessagesFromConversation = MessagesFromHistory.NONE }
       }
-      val model = CreateChatCompletionRequestModel.gpt_3_5_turbo_16k_0613
+      val model = CreateChatCompletionRequest.Model.Gpt35Turbo16k0613
       val config = AIConfig(model = model, openAI = openAI, conversation = this@Conversation)
       val animal: Animal =
         AI(
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/audio/SimpleSpeech.kt b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/audio/SimpleSpeech.kt
index 5a942d26b..5cec92a9b 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/audio/SimpleSpeech.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/audio/SimpleSpeech.kt
@@ -1,13 +1,13 @@
 package com.xebia.functional.xef.dsl.audio
 
 import com.sipgate.mp3wav.Converter
-import com.xebia.functional.openai.UploadFile
-import com.xebia.functional.openai.generated.model.CreateSpeechRequest
-import com.xebia.functional.openai.generated.model.CreateSpeechRequestModel
-import com.xebia.functional.openai.generated.model.CreateTranscriptionRequestModel
 import com.xebia.functional.xef.AI
 import com.xebia.functional.xef.Config
 import com.xebia.functional.xef.OpenAI
+import com.xebia.functional.xef.openapi.CreateSpeechRequest
+import com.xebia.functional.xef.openapi.CreateTranscriptionRequest
+import com.xebia.functional.xef.openapi.CreateTranslationRequest
+import com.xebia.functional.xef.openapi.UploadFile
 import io.ktor.client.statement.*
 import io.ktor.utils.io.jvm.javaio.*
 import java.io.File
@@ -28,11 +28,11 @@ suspend fun main() {
           .trimIndent()
       )
     val channel =
-      audio.createSpeech(
+      audio.speech.createSpeech(
         CreateSpeechRequest(
-          model = CreateSpeechRequestModel.tts_1,
+          model = CreateSpeechRequest.Model.Tts1,
           input = modelResponse,
-          voice = CreateSpeechRequest.Voice.nova
+          voice = CreateSpeechRequest.Voice.Nova
         )
       )
     val wavConverter = Converter(channel.bodyAsChannel().toInputStream())
@@ -51,19 +51,23 @@ suspend fun main() {
         bodyBuilder = { file.readBytes().forEach { writeByte(it) } }
       )
     val transcription =
-      audio.createTranscription(
-        model = CreateTranscriptionRequestModel.whisper_1,
-        prompt = "Translate to spanish",
-        file = uploadFile,
-        language = "es",
+      audio.transcriptions.createTranscription(
+        CreateTranscriptionRequest(
+          file = uploadFile,
+          model = CreateTranscriptionRequest.Model.Whisper1,
+          prompt = "Translate to spanish",
+          language = "es"
+        )
       )
     println("transcription in `es`:")
     println(transcription)
     val translation =
-      audio.createTranslation(
-        model = CreateTranscriptionRequestModel.whisper_1,
-        prompt = "Translate",
-        file = uploadFile,
+      audio.translations.createTranslation(
+        CreateTranslationRequest(
+          file = uploadFile,
+          model = CreateTranslationRequest.Model.Whisper1,
+          prompt = "Translate"
+        )
       )
     println("translation to english:")
     println(translation)
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/Classes.kt b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/Classes.kt
index 4842c3844..2cd82eac7 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/Classes.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/Classes.kt
@@ -1,11 +1,11 @@
 package com.xebia.functional.xef.dsl.chat
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.AI
 import com.xebia.functional.xef.conversation.Description
 import com.xebia.functional.xef.conversation.MessagePolicy
 import com.xebia.functional.xef.conversation.MessagesFromHistory
 import com.xebia.functional.xef.conversation.MessagesToHistory
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.assistant
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.system
@@ -26,7 +26,7 @@ data class Book(
 )
 
 suspend fun books(topic: String): Books {
-  val model = CreateChatCompletionRequestModel.gpt_4_turbo_preview
+  val model = CreateChatCompletionRequest.Model.Gpt4TurboPreview
 
   val myCustomPrompt =
     Prompt(
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/ParallelToolCallsAnonymous.kt b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/ParallelToolCallsAnonymous.kt
index 2cfdc904b..3a8027ebf 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/ParallelToolCallsAnonymous.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/chat/ParallelToolCallsAnonymous.kt
@@ -1,11 +1,11 @@
 package com.xebia.functional.xef.dsl.chat
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.AI
 import com.xebia.functional.xef.AIConfig
 import com.xebia.functional.xef.AIEvent
 import com.xebia.functional.xef.Tool
 import com.xebia.functional.xef.conversation.Description
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import kotlinx.coroutines.flow.Flow
 
 suspend fun ballLocationInfoFromLastCupTriedImpl(input: Int): String {
@@ -27,7 +27,7 @@ suspend fun main() {
       prompt = "Where is the ball? use the available tools to find out.",
       config =
         AIConfig(
-          model = CreateChatCompletionRequestModel.gpt_3_5_turbo,
+          model = CreateChatCompletionRequest.Model.Gpt35Turbo,
           tools =
             listOf(
               Tool.suspend(
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/images/ImagesExamples.kt b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/images/ImagesExamples.kt
index 01b71aa19..50c88ab74 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/dsl/images/ImagesExamples.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/dsl/images/ImagesExamples.kt
@@ -1,18 +1,20 @@
 package com.xebia.functional.xef.dsl.images
 
-import com.xebia.functional.openai.generated.model.CreateImageRequest
 import com.xebia.functional.xef.OpenAI
 import com.xebia.functional.xef.llm.asInputProvider
+import com.xebia.functional.xef.openapi.CreateImageRequest
+import com.xebia.functional.xef.openapi.CreateImageVariationRequest
+import com.xebia.functional.xef.openapi.OpenAI
 
 suspend fun main() {
   val openAI = OpenAI(logRequests = false)
   val ai = openAI.images
-  val image =
-    ai.createImage(
-      createImageRequest = CreateImageRequest(prompt = "Event horizon in a black hole")
-    )
+  val image = ai.generations.createImage(CreateImageRequest("Event horizon in a black hole"))
   val generatedImage = image.data.first()
   println("Image: $generatedImage")
-  val variant = ai.createImageVariation(generatedImage.asInputProvider())
+  val variant =
+    ai.variations.createImageVariation(
+      CreateImageVariationRequest(generatedImage.asInputProvider())
+    )
   println("Variant: $variant")
 }
diff --git a/examples/src/main/kotlin/com/xebia/functional/xef/evaluator/TestExample.kt b/examples/src/main/kotlin/com/xebia/functional/xef/evaluator/TestExample.kt
index e51f70487..67a40d588 100644
--- a/examples/src/main/kotlin/com/xebia/functional/xef/evaluator/TestExample.kt
+++ b/examples/src/main/kotlin/com/xebia/functional/xef/evaluator/TestExample.kt
@@ -1,7 +1,6 @@
 package com.xebia.functional.xef.evaluator
 
 import arrow.continuations.SuspendApp
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
 import com.xebia.functional.xef.OpenAI
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.evaluator.metrics.AnswerAccuracy
@@ -9,6 +8,8 @@ import com.xebia.functional.xef.evaluator.models.ModelsPricing
 import com.xebia.functional.xef.evaluator.models.OutputDescription
 import com.xebia.functional.xef.evaluator.models.OutputResponse
 import com.xebia.functional.xef.llm.promptMessageAndUsage
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
+import com.xebia.functional.xef.openapi.OpenAI
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.user
 import java.io.File
@@ -17,13 +18,13 @@ object TestExample {
 
   @JvmStatic
   fun main(args: Array<String>) = SuspendApp {
-    val model = CreateChatCompletionRequestModel.gpt_3_5_turbo_16k
+    val model = CreateChatCompletionRequest.Model.Gpt35Turbo16k
     val chat = OpenAI(logRequests = false).chat
 
     val spec =
       SuiteSpec(
         description = "Check GTP3.5 and fake outputs",
-        model = CreateChatCompletionRequestModel.gpt_4_turbo_preview
+        model = CreateChatCompletionRequest.Model.Gpt4TurboPreview
       ) {
         val gpt35Description = OutputDescription("Using GPT3.5")
         val fakeOutputs = OutputDescription("Fake outputs with errors")
diff --git a/gradle/libs.versions.toml b/gradle/libs.versions.toml
index 8f2d20937..12b6707de 100644
--- a/gradle/libs.versions.toml
+++ b/gradle/libs.versions.toml
@@ -39,7 +39,7 @@ progressbar = "0.10.0"
 jmf = "2.1.1e"
 mp3-wav-converter = "1.0.4"
 yamlkt="0.13.0"
-
+openapi-kt="0.0.6"
 
 [libraries]
 arrow-core = { module = "io.arrow-kt:arrow-core", version.ref = "arrow" }
@@ -143,3 +143,4 @@ semver-gradle = { id="com.javiersc.semver", version.ref="semverGradle" }
 resources = { id="com.goncalossilva.resources", version.ref="resources-kmp" }
 detekt = { id="io.gitlab.arturbosch.detekt", version.ref="detekt"}
 node-gradle = { id = "com.github.node-gradle.node", version.ref = "node-gradle" }
+openapi-kt = { id= "io.github.nomisrev.openapi-kt-plugin", version.ref = "openapi-kt" }
diff --git a/integrations/gcp/src/commonMain/kotlin/com/xebia/functional/xef/gcp/GcpClient.kt b/integrations/gcp/src/commonMain/kotlin/com/xebia/functional/xef/gcp/GcpClient.kt
index b668193fd..bc3d343e4 100644
--- a/integrations/gcp/src/commonMain/kotlin/com/xebia/functional/xef/gcp/GcpClient.kt
+++ b/integrations/gcp/src/commonMain/kotlin/com/xebia/functional/xef/gcp/GcpClient.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.gcp
 
-import com.xebia.functional.openai.apis.EmbeddingsApi
-import com.xebia.functional.openai.models.CreateEmbeddingRequest
-import com.xebia.functional.openai.models.ext.embedding.create.CreateEmbeddingRequestInput
+import com.xebia.functional.xef.openapi.apis.EmbeddingsApi
+import com.xebia.functional.xef.openapi.models.CreateEmbeddingRequest
+import com.xebia.functional.xef.openapi.models.ext.embedding.create.CreateEmbeddingRequestInput
 import com.xebia.functional.xef.AIError
 import com.xebia.functional.xef.conversation.AutoClose
 import com.xebia.functional.xef.conversation.autoClose
diff --git a/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryAssistantState.kt b/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryAssistantState.kt
index 616c17a75..059c94f77 100644
--- a/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryAssistantState.kt
+++ b/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryAssistantState.kt
@@ -1,6 +1,6 @@
 package com.xebia.functional.xef.opentelemetry
 
-import com.xebia.functional.openai.generated.model.*
+import com.xebia.functional.xef.openapi.*
 import io.opentelemetry.api.trace.Span
 import io.opentelemetry.api.trace.SpanKind
 import io.opentelemetry.api.trace.Tracer
@@ -146,23 +146,23 @@ class OpenTelemetryAssistantState(private val tracer: Tracer) {
     span.setAttribute("openai.assistant.runStep.id", id)
     span.setAttribute("openai.assistant.status", status.name)
     when (val detail = stepDetails) {
-      is RunStepObjectStepDetails.CaseRunStepDetailsMessageCreationObject -> {
+      is RunStepObject.StepDetails.CaseRunStepDetailsMessageCreationObject -> {
         span.setAttribute(
           "openai.assistant.messageCreation.id",
           detail.value.messageCreation.messageId
         )
       }
-      is RunStepObjectStepDetails.CaseRunStepDetailsToolCallsObject -> {
+      is RunStepObject.StepDetails.CaseRunStepDetailsToolCallsObject -> {
         detail.value.toolCalls.forEachIndexed { index, toolCall ->
           when (toolCall) {
-            is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsCodeObject -> {
+            is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsCodeObject -> {
               span.setAttribute("openai.assistant.toolCalls.$index.type", toolCall.value.type.name)
               span.setAttribute(
                 "openai.assistant.toolCalls.$index.function.name",
                 "code_interpreter"
               )
             }
-            is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsFunctionObject -> {
+            is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsFunctionObject -> {
               span.setAttribute("openai.assistant.toolCalls.$index.type", toolCall.value.type.name)
               span.setAttribute(
                 "openai.assistant.toolCalls.$index.function.name",
@@ -173,7 +173,7 @@ class OpenTelemetryAssistantState(private val tracer: Tracer) {
                 toolCall.value.function.arguments ?: ""
               )
             }
-            is RunStepDetailsToolCallsObjectToolCallsInner.CaseRunStepDetailsToolCallsFileSearchObject -> {
+            is RunStepDetailsToolCallsObject.ToolCalls.CaseRunStepDetailsToolCallsFileSearchObject -> {
               span.setAttribute("openai.assistant.toolCalls.$index.type", toolCall.value.type.name)
               span.setAttribute("openai.assistant.toolCalls.$index.function.name", "retrieval")
             }
@@ -192,13 +192,13 @@ class OpenTelemetryAssistantState(private val tracer: Tracer) {
     span.setAttribute("openai.assistant.message.id", id)
     status?.let { span.setAttribute("openai.assistant.message.status", it.name) }
     when (val inner = content.firstOrNull()) {
-      is MessageObjectContentInner.CaseMessageContentImageFileObject -> {
+      is MessageObject.Content.CaseMessageContentImageFileObject -> {
         span.setAttribute("openai.assistant.message.content", inner.value.imageFile.fileId)
       }
-      is MessageObjectContentInner.CaseMessageContentTextObject -> {
+      is MessageObject.Content.CaseMessageContentTextObject -> {
         span.setAttribute("openai.assistant.message.content", inner.value.text.value)
       }
-      is MessageObjectContentInner.CaseMessageContentImageUrlObject ->
+      is MessageObject.Content.CaseMessageContentImageUrlObject ->
         span.setAttribute("openai.assistant.message.content", inner.value.imageUrl.url)
       null -> {}
     }
diff --git a/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryMetric.kt b/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryMetric.kt
index 34f859818..a5dc754f9 100644
--- a/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryMetric.kt
+++ b/integrations/opentelemetry/src/main/kotlin/com/xebia/functional/xef/opentelemetry/OpenTelemetryMetric.kt
@@ -1,9 +1,9 @@
 package com.xebia.functional.xef.opentelemetry
 
-import com.xebia.functional.openai.generated.model.MessageObject
-import com.xebia.functional.openai.generated.model.RunObject
-import com.xebia.functional.openai.generated.model.RunStepObject
 import com.xebia.functional.xef.metrics.Metric
+import com.xebia.functional.xef.openapi.MessageObject
+import com.xebia.functional.xef.openapi.RunObject
+import com.xebia.functional.xef.openapi.RunStepObject
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.contentAsString
 import io.opentelemetry.api.trace.*
diff --git a/integrations/postgresql/src/main/kotlin/com/xebia/functional/xef/store/PostgreSQLVectorStore.kt b/integrations/postgresql/src/main/kotlin/com/xebia/functional/xef/store/PostgreSQLVectorStore.kt
index cd32df067..1e82b8287 100644
--- a/integrations/postgresql/src/main/kotlin/com/xebia/functional/xef/store/PostgreSQLVectorStore.kt
+++ b/integrations/postgresql/src/main/kotlin/com/xebia/functional/xef/store/PostgreSQLVectorStore.kt
@@ -1,16 +1,12 @@
 package com.xebia.functional.xef.store
 
 import arrow.atomic.AtomicInt
-import com.xebia.functional.openai.generated.api.Embeddings
-import com.xebia.functional.openai.generated.model.ChatCompletionRole
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
 import com.xebia.functional.xef.llm.embedDocuments
 import com.xebia.functional.xef.llm.embedQuery
 import com.xebia.functional.xef.llm.models.modelType
 import com.xebia.functional.xef.prompt.contentAsString
 import com.xebia.functional.xef.store.postgresql.*
+import com.xebia.functional.xef.openapi.*
 import kotlinx.uuid.UUID
 import kotlinx.uuid.generateUUID
 import javax.sql.DataSource
@@ -22,7 +18,7 @@ class PGVectorStore(
   private val collectionName: String,
   private val distanceStrategy: PGDistanceStrategy,
   private val preDeleteCollection: Boolean,
-  private val embeddingRequestModel: CreateEmbeddingRequestModel,
+  private val embeddingRequestModel: CreateEmbeddingRequest.Model,
   private val chunkSize: Int = 400
 ) : VectorStore {
 
@@ -38,7 +34,7 @@ class PGVectorStore(
         update(addNewMemory) {
           bind(UUID.generateUUID().toString())
           bind(memory.conversationId.value)
-          bind(memory.content.role.value.lowercase())
+          bind(memory.content.role.name)
           bind(memory.content.asRequestMessage().contentAsString())
           bind(memory.index)
         }
@@ -46,7 +42,7 @@ class PGVectorStore(
     }
   }
 
-  override suspend fun memories(model: CreateChatCompletionRequestModel, conversationId: ConversationId, limitTokens: Int): List<Memory> =
+  override suspend fun memories(model: CreateChatCompletionRequest.Model, conversationId: ConversationId, limitTokens: Int): List<Memory> =
     getMemoryByConversationId(conversationId).reduceByLimitToken(model.modelType(), limitTokens).reversed()
 
   private fun JDBCSyntax.getCollection(collectionName: String): PGCollection =
@@ -153,7 +149,7 @@ class PGVectorStore(
         val index = int()
         Memory(
           conversationId = ConversationId(cId),
-          content = memorizedMessage(ChatCompletionRole.valueOf(role.lowercase()), content),
+          content = memorizedMessage(ChatCompletionRole.valueOf(role), content),
           index = index
         )
       }
diff --git a/integrations/postgresql/src/test/kotlin/xef/MemoryData.kt b/integrations/postgresql/src/test/kotlin/xef/MemoryData.kt
index 00e7a76b9..541c42772 100644
--- a/integrations/postgresql/src/test/kotlin/xef/MemoryData.kt
+++ b/integrations/postgresql/src/test/kotlin/xef/MemoryData.kt
@@ -1,10 +1,9 @@
 package xef
 
 import arrow.atomic.AtomicInt
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestAssistantMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestUserMessage
-import com.xebia.functional.openai.generated.model.ChatCompletionRequestUserMessageContent
+import com.xebia.functional.xef.openapi.ChatCompletionRequestAssistantMessage
+import com.xebia.functional.xef.openapi.ChatCompletionRequestMessage
+import com.xebia.functional.xef.openapi.ChatCompletionRequestUserMessage
 import com.xebia.functional.xef.store.ConversationId
 import com.xebia.functional.xef.store.MemorizedMessage
 import com.xebia.functional.xef.store.Memory
@@ -22,13 +21,13 @@ class MemoryData {
         (0 until n).flatMap {
             val m1 = ChatCompletionRequestMessage.CaseChatCompletionRequestUserMessage(
                 ChatCompletionRequestUserMessage(
-                    role = ChatCompletionRequestUserMessage.Role.user,
-                    content = ChatCompletionRequestUserMessageContent.CaseString("Question $it${append?.let { ": $it" } ?: ""}")
+                    role = ChatCompletionRequestUserMessage.Role.User,
+                    content = ChatCompletionRequestUserMessage.Content.CaseString("Question $it${append?.let { ": $it" } ?: ""}")
                 )
             )
             val m2 = ChatCompletionRequestMessage.CaseChatCompletionRequestAssistantMessage(
                 ChatCompletionRequestAssistantMessage(
-                    role = ChatCompletionRequestAssistantMessage.Role.assistant,
+                    role = ChatCompletionRequestAssistantMessage.Role.Assistant,
                     content = "Answer $it${append?.let { ": $it" } ?: ""}"
                 )
             )
diff --git a/integrations/postgresql/src/test/kotlin/xef/PGVectorStoreSpec.kt b/integrations/postgresql/src/test/kotlin/xef/PGVectorStoreSpec.kt
index c7e5ab83b..f916e26de 100644
--- a/integrations/postgresql/src/test/kotlin/xef/PGVectorStoreSpec.kt
+++ b/integrations/postgresql/src/test/kotlin/xef/PGVectorStoreSpec.kt
@@ -1,13 +1,11 @@
 package xef
 
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
-import com.xebia.functional.openai.generated.model.Embedding
 import com.xebia.functional.xef.store.PGVectorStore
 import com.xebia.functional.xef.store.migrations.runDatabaseMigrations
 import com.xebia.functional.xef.store.postgresql.PGDistanceStrategy
 import com.zaxxer.hikari.HikariConfig
 import com.zaxxer.hikari.HikariDataSource
+import com.xebia.functional.xef.openapi.*
 import io.kotest.core.Tuple3
 import io.kotest.core.extensions.install
 import io.kotest.core.spec.style.StringSpec
@@ -39,7 +37,7 @@ class PGVectorStoreSpec :
         )
       )
 
-    val embeddingsRequestModel = CreateEmbeddingRequestModel.text_embedding_ada_002
+    val embeddingsRequestModel = CreateEmbeddingRequest.Model.TextEmbeddingAda002
 
     fun StringSpecScope.pg() =
       PGVectorStore(
@@ -87,7 +85,7 @@ class PGVectorStoreSpec :
 
     "similaritySearchByVector should return both documents" {
       pg().addTexts(listOf("bar", "foo"))
-      pg().similaritySearchByVector(Embedding(0, listOf(4.0, 5.0, 6.0), Embedding.Object.embedding), 2) shouldBe
+      pg().similaritySearchByVector(Embedding(0, listOf(4.0, 5.0, 6.0), Embedding.Object.Embedding), 2) shouldBe
         listOf("bar", "foo")
     }
 
@@ -102,14 +100,14 @@ class PGVectorStoreSpec :
 
     "similaritySearchByVector should return document" {
       pg().similaritySearchByVector(
-        Embedding(0, listOf(1.0, 2.0, 3.0), Embedding.Object.embedding),
+        Embedding(0, listOf(1.0, 2.0, 3.0), Embedding.Object.Embedding),
         1
       ) shouldBe listOf("foo")
     }
 
     "the added memories sorted by index should be obtained in the same order" {
       val memoryData = MemoryData()
-      val model = CreateChatCompletionRequestModel.gpt_4
+      val model = CreateChatCompletionRequest.Model.Gpt4
       val memories = memoryData.generateRandomMessages(10)
       pg().addMemories(memories)
       memories.map { Tuple3(it.index, it.conversationId, it.content.asRequestMessage()) } shouldBe
diff --git a/integrations/postgresql/src/test/kotlin/xef/TestEmbeddings.kt b/integrations/postgresql/src/test/kotlin/xef/TestEmbeddings.kt
index a56584375..3b614e49d 100644
--- a/integrations/postgresql/src/test/kotlin/xef/TestEmbeddings.kt
+++ b/integrations/postgresql/src/test/kotlin/xef/TestEmbeddings.kt
@@ -1,7 +1,6 @@
 package xef
 
-import com.xebia.functional.openai.generated.api.Embeddings
-import com.xebia.functional.openai.generated.model.*
+import com.xebia.functional.xef.openapi.*
 import io.ktor.client.request.*
 
 class TestEmbeddings : Embeddings, AutoCloseable {
@@ -14,25 +13,25 @@ class TestEmbeddings : Embeddings, AutoCloseable {
   ): CreateEmbeddingResponse {
     requests.add(createEmbeddingRequest)
     val maybeTextInput = when(val input = createEmbeddingRequest.input) {
-      is CreateEmbeddingRequestInput.CaseStrings -> input.value.firstOrNull()
-      is CreateEmbeddingRequestInput.CaseString -> input.value
+      is CreateEmbeddingRequest.Input.CaseStrings -> input.value.firstOrNull()
+      is CreateEmbeddingRequest.Input.CaseString -> input.value
       else -> null
     }
     val data = when(maybeTextInput) {
       "fooz" -> listOf(
-        Embedding(0, listOf(7.0, 8.0, 9.0), Embedding.Object.embedding),
-        Embedding(1, listOf(7.5, 8.5, 9.5), Embedding.Object.embedding)
+        Embedding(0, listOf(7.0, 8.0, 9.0), Embedding.Object.Embedding),
+        Embedding(1, listOf(7.5, 8.5, 9.5), Embedding.Object.Embedding)
       )
-      "foo" -> listOf(Embedding(0, listOf(1.0, 2.0, 3.0), Embedding.Object.embedding))
-      "bar" -> listOf(Embedding(0, listOf(4.0, 5.0, 6.0), Embedding.Object.embedding))
+      "foo" -> listOf(Embedding(0, listOf(1.0, 2.0, 3.0), Embedding.Object.Embedding))
+      "bar" -> listOf(Embedding(0, listOf(4.0, 5.0, 6.0), Embedding.Object.Embedding))
       "baz" -> listOf()
       else -> listOf()
     }
     return CreateEmbeddingResponse(
       data = data,
       model = "test-model",
-      `object` = CreateEmbeddingResponse.Object.list,
-      usage = CreateEmbeddingResponseUsage(0, 0)
+      `object` = CreateEmbeddingResponse.Object.List,
+      usage = CreateEmbeddingResponse.Usage(0, 0)
     )
   }
 
diff --git a/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/SQL.kt b/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/SQL.kt
index 6aceaf222..869ec65ff 100644
--- a/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/SQL.kt
+++ b/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/SQL.kt
@@ -1,12 +1,12 @@
 package com.xebia.functional.xef.sql
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
+import com.xebia.functional.xef.openapi.Chat
 import com.xebia.functional.xef.Tool
 import com.xebia.functional.xef.conversation.AiDsl
 import com.xebia.functional.xef.conversation.Conversation
 import com.xebia.functional.xef.conversation.Description
 import com.xebia.functional.xef.llm.*
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 import com.xebia.functional.xef.prompt.Prompt
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.system
 import com.xebia.functional.xef.prompt.PromptBuilder.Companion.user
@@ -44,7 +44,7 @@ interface SQL {
     suspend fun Conversation.promptQuery(prompt: String, tableNames: List<String>, context: String?): AnswerResponse
 }
 
-class SQLImpl(private val chatApi: Chat, private val model: CreateChatCompletionRequestModel, private val db: Database) : SQL {
+class SQLImpl(private val chatApi: Chat, private val model: CreateChatCompletionRequest.Model, private val db: Database) : SQL {
     private val logger = KotlinLogging.logger {}
 
     override suspend fun Conversation.promptQuery(
diff --git a/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/jdbc/JdbcConfig.kt b/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/jdbc/JdbcConfig.kt
index c740a3dc5..ac5e56193 100644
--- a/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/jdbc/JdbcConfig.kt
+++ b/integrations/sql/src/main/kotlin/com/xebia/functional/xef/sql/jdbc/JdbcConfig.kt
@@ -1,7 +1,7 @@
 package com.xebia.functional.xef.sql.jdbc
 
-import com.xebia.functional.openai.generated.api.Chat
-import com.xebia.functional.openai.generated.model.CreateChatCompletionRequestModel
+import com.xebia.functional.xef.openapi.Chat
+import com.xebia.functional.xef.openapi.CreateChatCompletionRequest
 
 class JdbcConfig(
   val vendor: String,
@@ -11,7 +11,7 @@ class JdbcConfig(
   val port: Int,
   val database: String,
   val chatApi: Chat,
-  val model: CreateChatCompletionRequestModel
+  val model: CreateChatCompletionRequest.Model
 ) {
   fun toJDBCUrl(): String = "jdbc:$vendor://$host:$port/$database"
 }
diff --git a/openai-client/client/.openapi-generator-ignore b/openai-client/client/.openapi-generator-ignore
deleted file mode 100644
index 4b8839f00..000000000
--- a/openai-client/client/.openapi-generator-ignore
+++ /dev/null
@@ -1,15 +0,0 @@
-.gitignore
-build.gradle.kts
-settings.gradle.kts
-gradle/**
-gradlew
-gradlew.bat
-README.md
-docs/**
-
-build.gradle
-settings.gradle
-
-**/src/test/
-/**/org/openapitools/client/infrastructure/
-**/proguard-rules.pro
diff --git a/openai-client/client/build.gradle.kts b/openai-client/client/build.gradle.kts
deleted file mode 100644
index 7beb5a2a1..000000000
--- a/openai-client/client/build.gradle.kts
+++ /dev/null
@@ -1,138 +0,0 @@
-@file:Suppress("DSL_SCOPE_VIOLATION")
-
-import org.jetbrains.dokka.gradle.DokkaTask
-
-repositories { mavenCentral() }
-
-plugins {
-    base
-    alias(libs.plugins.kotlin.multiplatform)
-    alias(libs.plugins.kotest.multiplatform)
-    alias(libs.plugins.kotlinx.serialization)
-    alias(libs.plugins.spotless)
-    alias(libs.plugins.dokka)
-    alias(libs.plugins.arrow.gradle.publish)
-    alias(libs.plugins.semver.gradle)
-    alias(libs.plugins.detekt)
-}
-
-dependencies { detektPlugins(project(":detekt-rules")) }
-
-java {
-    sourceCompatibility = JavaVersion.VERSION_11
-    targetCompatibility = JavaVersion.VERSION_11
-    toolchain { languageVersion = JavaLanguageVersion.of(11) }
-}
-
-kotlin {
-    sourceSets.commonMain {
-        kotlin.srcDir(project.file("build/generated/OpenAI/src/commonMain/kotlin"))
-    }
-}
-
-// Automatically run the generator when importing
-tasks.maybeCreate("prepareKotlinIdeaImport")
-    .dependsOn(":xef-openai-client-generator:openaiClientGenerate")
-
-detekt {
-    toolVersion = "1.23.1"
-    source.setFrom("src/commonMain/kotlin", "src/jvmMain/kotlin")
-    config.setFrom("../../config/detekt/detekt.yml")
-    autoCorrect = true
-}
-
-kotlin {
-    jvm()
-    js(IR) {
-        browser()
-        nodejs()
-        // TODO support wasm, etc
-    }
-    linuxX64()
-    macosX64()
-    macosArm64()
-    mingwX64()
-    // iOS, Android, etc?
-    sourceSets {
-        all { languageSettings.optIn("kotlin.ExperimentalStdlibApi") }
-        val commonMain by getting {
-            dependencies {
-                api(projects.xefTokenizer)
-                implementation(libs.ktor.client.content.negotiation)
-                implementation(libs.ktor.client.serialization)
-                implementation(libs.ktor.client.logging)
-                implementation(libs.klogging)
-            }
-        }
-        val commonTest by getting {
-            dependencies {
-                implementation(kotlin("test"))
-                implementation(libs.kotest.property)
-                implementation(libs.kotest.assertions)
-            }
-        }
-        val jvmMain by getting {
-            dependencies {
-                api(libs.ktor.client.cio)
-            }
-        }
-        val jsMain by getting { dependencies { api(libs.ktor.client.js) } }
-        val jvmTest by getting { dependencies { implementation(libs.kotest.junit5) } }
-        val linuxX64Main by getting { dependencies { api(libs.ktor.client.cio) } }
-        val macosX64Main by getting { dependencies { api(libs.ktor.client.cio) } }
-        val macosArm64Main by getting { dependencies { api(libs.ktor.client.cio) } }
-        val mingwX64Main by getting { dependencies { api(libs.ktor.client.winhttp) } }
-    }
-}
-
-spotless {
-    kotlin {
-        target("**/*.kt")
-        ktfmt().googleStyle().configure { it.setRemoveUnusedImport(true) }
-    }
-}
-
-tasks {
-    withType<io.gitlab.arturbosch.detekt.Detekt>().configureEach {
-        dependsOn(":detekt-rules:assemble")
-        autoCorrect = true
-    }
-    named("detektJvmMain") {
-        dependsOn(":detekt-rules:assemble")
-        getByName("build").dependsOn(this)
-    }
-    named("detekt") {
-        dependsOn(":detekt-rules:assemble")
-        getByName("build").dependsOn(this)
-    }
-    withType<Test>().configureEach {
-        maxParallelForks = Runtime.getRuntime().availableProcessors()
-        useJUnitPlatform()
-        testLogging {
-            setExceptionFormat("full")
-            setEvents(listOf("passed", "skipped", "failed", "standardOut", "standardError"))
-        }
-    }
-    withType<DokkaTask>().configureEach {
-        kotlin.sourceSets.forEach { kotlinSourceSet ->
-            dokkaSourceSets.named(kotlinSourceSet.name) {
-                perPackageOption {
-                    matchingRegex.set(".*\\.internal.*")
-                    suppress.set(true)
-                }
-                skipDeprecated.set(true)
-                reportUndocumented.set(false)
-                val baseUrl = checkNotNull(project.properties["pom.smc.url"]?.toString())
-                kotlinSourceSet.kotlin.srcDirs.filter { it.exists() }.forEach { srcDir ->
-                    sourceLink {
-                        localDirectory.set(srcDir)
-                        remoteUrl.set(uri("$baseUrl/blob/main/${srcDir.relativeTo(rootProject.rootDir)}").toURL())
-                        remoteLineSuffix.set("#L")
-                    }
-                }
-            }
-        }
-    }
-}
-
-tasks.withType<AbstractPublishToMaven> { dependsOn(tasks.withType<Sign>()) }
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/AssistantEvent.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/AssistantEvent.kt
deleted file mode 100644
index 64f9e266d..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/AssistantEvent.kt
+++ /dev/null
@@ -1,8 +0,0 @@
-package com.xebia.functional.openai
-
-import kotlinx.serialization.Serializable
-
-// TODO write proper AssistantEvent
-// According to:
-// https://platform.openai.com/docs/api-reference/assistants-streaming/message-delta-object
-@Serializable data class AssistantEvent(val todo: String)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/Config.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/Config.kt
deleted file mode 100644
index 3efb3c9b9..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/Config.kt
+++ /dev/null
@@ -1,12 +0,0 @@
-package com.xebia.functional.openai
-
-import kotlinx.serialization.json.Json
-
-data class Config(
-  val baseUrl: String,
-  val token: String,
-  val org: String?,
-  val json: Json,
-  val streamingPrefix: String,
-  val streamingDelimiter: String
-)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/FormBuilderOps.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/FormBuilderOps.kt
deleted file mode 100644
index 13fed8050..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/FormBuilderOps.kt
+++ /dev/null
@@ -1,39 +0,0 @@
-package com.xebia.functional.openai
-
-import io.ktor.client.request.forms.*
-import io.ktor.http.*
-import io.ktor.util.*
-import io.ktor.utils.io.core.*
-import kotlinx.serialization.ExperimentalSerializationApi
-import kotlinx.serialization.InternalSerializationApi
-import kotlinx.serialization.serializerOrNull
-
-@OptIn(InternalAPI::class)
-fun <T : Any> FormBuilder.appendGen(key: String, value: T, headers: Headers = Headers.Empty): Unit {
-  when (value) {
-    is String -> append(key, value, headers)
-    is Number -> append(key, value, headers)
-    is Boolean -> append(key, value, headers)
-    is ByteArray -> append(key, value, headers)
-    is ByteReadPacket -> append(key, value, headers)
-    is InputProvider -> append(key, value, headers)
-    is ChannelProvider -> append(key, value, headers)
-    is UploadFile -> appendUploadedFile(key, value)
-    is Enum<*> -> append(key, serialNameOrEnumValue(value), headers)
-    else -> append(key, value, headers)
-  }
-}
-
-internal fun FormBuilder.appendUploadedFile(key: String, file: UploadFile) {
-  append(
-    key = key,
-    filename = file.filename,
-    contentType = file.contentType ?: ContentType.Application.OctetStream,
-    size = file.size,
-    bodyBuilder = file.bodyBuilder
-  )
-}
-
-@OptIn(InternalSerializationApi::class, ExperimentalSerializationApi::class)
-fun <T : Enum<T>> serialNameOrEnumValue(v: Enum<T>): String =
-  v::class.serializerOrNull()?.descriptor?.getElementName(v.ordinal) ?: v.toString()
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/OneOfOps.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/OneOfOps.kt
deleted file mode 100644
index 0cd3ed386..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/OneOfOps.kt
+++ /dev/null
@@ -1,53 +0,0 @@
-package com.xebia.functional.openai
-
-import kotlin.reflect.KClass
-import kotlinx.serialization.SerializationException
-import kotlinx.serialization.json.JsonElement
-
-class OneOfSerializationException(
-  val payload: JsonElement,
-  val errors: Map<KClass<*>, SerializationException>,
-  override val message: String =
-    """
-    Failed to deserialize Json: $payload.
-    Errors: ${
-    errors.entries.joinToString(separator = "\n") { (type, error) ->
-      "$type - failed to deserialize: ${error.stackTraceToString()}"
-    }
-  }
-    """
-      .trimIndent()
-) : SerializationException(message)
-
-/**
- * OpenAI makes a lot of use of oneOf types (sum types, or unions types), but it **never** relies on
- * a discriminator field to differentiate between the types.
- *
- * Typically, what OpenAI does is attach a common field like `type` (a single value enum). I.e.
- * `MessageObjectContentInner` has a type field with `image` or `text`. Depending on the `type`
- * property, the other properties will be different.
- *
- * Due to the use of these fields, it **seems** there are no overlapping objects in the schema. So
- * to deserialize these types, we can try to deserialize each type and return the first one that
- * succeeds. In the case **all** fail, we throw [OneOfSerializationException] which includes all the
- * attempted types with their errors.
- *
- * This method relies on 'peeking', which is not possible in KotlinX Serialization. So to achieve
- * peeking, we first deserialize the raw Json to JsonElement, which safely consumes the buffer. And
- * then we can attempt to deserialize the JsonElement to the desired type, without breaking the
- * internal parser buffer.
- */
-internal fun <A> attemptDeserialize(
-  json: JsonElement,
-  vararg block: Pair<KClass<*>, (json: JsonElement) -> A>
-): A {
-  val errors = linkedMapOf<KClass<*>, SerializationException>()
-  block.forEach { (kclass, f) ->
-    try {
-      return f(json)
-    } catch (e: SerializationException) {
-      errors[kclass] = e
-    }
-  }
-  throw OneOfSerializationException(json, errors)
-}
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/ServerSentEvent.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/ServerSentEvent.kt
deleted file mode 100644
index a3427178b..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/ServerSentEvent.kt
+++ /dev/null
@@ -1,10 +0,0 @@
-package com.xebia.functional.openai
-
-import kotlinx.serialization.Serializable
-import kotlinx.serialization.json.JsonElement
-
-@Serializable
-data class ServerSentEvent(
-  val event: String? = null,
-  val data: JsonElement? = null,
-)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/StreamingOps.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/StreamingOps.kt
deleted file mode 100644
index 6727ce030..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/StreamingOps.kt
+++ /dev/null
@@ -1,59 +0,0 @@
-package com.xebia.functional.openai
-
-import io.ktor.client.statement.HttpResponse
-import io.ktor.client.statement.bodyAsChannel
-import io.ktor.utils.io.ByteReadChannel
-import io.ktor.utils.io.readUTF8Line
-import kotlinx.coroutines.flow.FlowCollector
-import kotlinx.serialization.json.Json
-import kotlinx.serialization.json.JsonObject
-import kotlinx.serialization.serializer
-
-// ServerSentEvent, INTERNAL TO THIS MODULE
-// RunDelta | CreateChatCompletionStreamResponse
-internal suspend inline fun <reified A> FlowCollector<A>.streamEvents(
-  response: HttpResponse,
-  json: Json,
-  prefix: String,
-  end: String
-) {
-  val channel: ByteReadChannel = response.bodyAsChannel()
-  var nextEvent: String? = null
-  while (!channel.isClosedForRead) {
-    val line = channel.readUTF8Line() ?: continue
-
-    // break when we reach the end of the stream
-    if (line.startsWith(end)) {
-      break
-    }
-
-    // if the line is an event like "event: thread.created" we want to ensure
-    // A is of type ServerSentEvent and we skip the line treating next `prefix` as a JsonObject
-    // otherwise we treat the line as a json object if it starts with the prefix
-    // and emit the value
-
-    // if the line is an event like "event: thread.created" we want to ensure
-    // A is of type ServerSentEvent and we skip the line treating next `prefix` as a JsonObject
-    if (line.startsWith("event:")) {
-      nextEvent = line.removePrefix("event:").trim()
-      continue
-    }
-    // otherwise we treat the line as a json object if it starts with the prefix
-    else if (line.startsWith(prefix) && nextEvent == null) {
-      val data = line.removePrefix(prefix).trim()
-      val value: A = json.decodeFromString(serializer(), data)
-      emit(value)
-    }
-    // emit the value for the next event
-    if (nextEvent != null) {
-      val data = line.removePrefix(prefix).trim()
-      if (data.isNotBlank()) {
-        val eventData = json.decodeFromString(JsonObject.serializer(), data)
-        val value: A = ServerSentEvent(event = nextEvent, data = eventData) as A
-        emit(value)
-      }
-    }
-  }
-
-  // serverevent as A
-}
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/UploadFile.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/UploadFile.kt
deleted file mode 100644
index d97aaac84..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/UploadFile.kt
+++ /dev/null
@@ -1,11 +0,0 @@
-package com.xebia.functional.openai
-
-import io.ktor.http.*
-import io.ktor.utils.io.core.*
-
-data class UploadFile(
-  val filename: String,
-  val contentType: ContentType? = null,
-  val size: Long? = null,
-  val bodyBuilder: BytePacketBuilder.() -> Unit
-)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerError.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerError.kt
deleted file mode 100644
index c55a8e3aa..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerError.kt
+++ /dev/null
@@ -1,4 +0,0 @@
-package com.xebia.functional.openai.errors
-
-class ResponseSerializerError(message: String, cause: Throwable? = null) :
-  Exception(message, cause)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerErrors.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerErrors.kt
deleted file mode 100644
index 4fafce148..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/errors/ResponseSerializerErrors.kt
+++ /dev/null
@@ -1,33 +0,0 @@
-package com.xebia.functional.openai.errors
-
-import io.ktor.client.call.*
-import io.ktor.client.statement.*
-import io.ktor.http.content.*
-
-suspend inline fun <reified A> HttpResponse.serializeOrThrowWithResponseInfo(): A =
-  try {
-    this.body() ?: throw ResponseSerializerError("Response body is null")
-  } catch (e: Exception) {
-    val requestBody =
-      when (val content = this.request.content) {
-        is OutgoingContent.ByteArrayContent -> content.bytes().decodeToString()
-        is OutgoingContent.NoContent -> "NoContent"
-        is OutgoingContent.ProtocolUpgrade -> "ProtocolUpgrade"
-        is OutgoingContent.ReadChannelContent -> "ReadChannelContent"
-        is OutgoingContent.WriteChannelContent -> "WriteChannelContent"
-        else -> "UnknownContent"
-      }
-    throw ResponseSerializerError(
-      """
-      |Failed to serialize response body to ${A::class.simpleName}
-      |Request URL: ${this.request.url}
-      |Request Method: ${this.request.method}
-      |Request Body: $requestBody
-      |Response Status: ${this.status}
-      |Response Headers: ${this.headers}
-      |Response Body: ${this.bodyAsText()}
-    """
-        .trimMargin(),
-      e
-    )
-  }
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateAssistantRequestToolResourcesFileSearch.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateAssistantRequestToolResourcesFileSearch.kt
deleted file mode 100644
index ad9af6f21..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateAssistantRequestToolResourcesFileSearch.kt
+++ /dev/null
@@ -1,9 +0,0 @@
-package com.xebia.functional.openai.generated.model
-
-import kotlinx.serialization.SerialName
-import kotlinx.serialization.Serializable
-
-@Serializable
-data class CreateAssistantRequestToolResourcesFileSearch(
-  @SerialName(value = "vector_store_ids") val vectorStoreIds: List<String>
-)
diff --git a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateThreadRequestToolResourcesFileSearch.kt b/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateThreadRequestToolResourcesFileSearch.kt
deleted file mode 100644
index cf348e230..000000000
--- a/openai-client/client/src/commonMain/kotlin/com/xebia/functional/openai/generated/model/CreateThreadRequestToolResourcesFileSearch.kt
+++ /dev/null
@@ -1,9 +0,0 @@
-package com.xebia.functional.openai.generated.model
-
-import kotlinx.serialization.SerialName
-import kotlinx.serialization.Serializable
-
-@Serializable
-data class CreateThreadRequestToolResourcesFileSearch(
-  @SerialName(value = "vector_store_ids") val vectorStoreIds: List<String>
-)
diff --git a/openai-client/generator/.openapi-generator-ignore b/openai-client/generator/.openapi-generator-ignore
deleted file mode 100644
index dc1bce6c5..000000000
--- a/openai-client/generator/.openapi-generator-ignore
+++ /dev/null
@@ -1,9 +0,0 @@
-**/docs/**
-**/README.md
-**/test/**
-**/org/openapitools/**
-**/proguard-rules.pro
-**/build.gradle
-**/settings.gradle
-**/CreateAssistantRequestToolResourcesFileSearch.kt
-**/CreateThreadRequestToolResourcesFileSearch.kt
diff --git a/openai-client/generator/README.md b/openai-client/generator/README.md
deleted file mode 100644
index 97e0f75c0..000000000
--- a/openai-client/generator/README.md
+++ /dev/null
@@ -1,77 +0,0 @@
-# OpenAI Client
-
-Autogenerated client for OpenAI using the OpenAPI specification and [OpenAPI Generator](https://openapi-generator.tech/)
-
-## Download API Specification
-
-Requirements:
-* File `openai-client/generator/config/openai-api-commit` with the desired commit of the OpenAI API
-
-Running the task `downloadOpenAIAPI` will read that commit and download the specification into `openai-client/generator/config/openai-api.yaml`
-
-```shell
-./gradlew downloadOpenAIAPI
-```
-
-## Generate code
-
-Requirements:
-* File `openai-client/generator/config/openai-api.yaml` with the API specification
-
-Running the task `openaiClientGenerate` will:
-* Generate the code with OpenAI Generator using the configuration
-* Run the task `spotlessApply`
-
-```shell
-./gradlew openaiClientGenerate
-```
-
-## Further Customization
-
-We choose for a semi-manual approach to generate the code.
-With a preference for a more controlled way of generating the code, so prefer custom template over workarounds.
-This generator add support for `oneOf`, and potentially more in the future.
-
-Since we focus on KMP, we use KotlinX Serialization and Ktor.
-So by being opinionated, we can heavily simply the templates. Which is great for maintainability.
-
-- [KMMGenerator](src/main/kotlin/com/openai/generator/KMMGenerator.kt) that generates the code.
-- config/*.mustache files with the templates for the code generation.
-  - api.mustache: invoked for each API endpoint. Generates interface, and smart-constructor.
-  - openai.mustache: the OpenAI interface that exposes all the API endpoints, with fluent syntax.
-  - model.mustache: invoked for each model. Delegate to the model templates (data, enum, oneOf). 
-    - data.mustache: generates the data class for the model (including nested models).
-    - enum.mustache: generates the enum class for the model (including nested models).
-    - oneOf.mustache: generates the sealed class for the model (including nested models).
-
-### Debugging
-
-```
-./gradlew openaiClientGenerate -PdebugModels=true > model.log
-```
-```
-./gradlew openaiClientGenerate -PdebugOperations=true > operations.log
-```
-
-You can only debug one of the two properties at a time. _operations_ are used to debug the API endpoints, _models_ are used to debug the models.
-It's best to dump the output in a file, such that you can inspect them in a more controlled way,
-but they should be included in git!
-
-### Mustache
-
-[Mustache(5)](https://mustache.github.io) is being used as template engine,
-with [Mustache Java](https://github.com/spullara/mustache.java).
-
-Quick reference:
-  - triple `{`: {{{var_name}}} is (HTML escaped)
-  - double `{{`: {{var_name}} is plain value
-  - {{#var_name}} ... {{/var_name}} is a section, drill down into model, for each, value present, etc.
-  - {{^var_name}} ... {{/var_name}} is a negative section, if var_name is empty, or false, etc.
-  - {{! comment }} is a comment
-  - {{> partial }} is a partial, include another template
-  - {{& var_name }} is a triple mustache without escaping
-  - {{#-first}} ... {{/-first}} only execute for first/last element in an array
-  - {{^-last}} ... {{/-last}} only execute for first/last element in an array
-  - {{#lambda.name}} ... {{/lambda.name}} is a lambda, a function that can be called with a parameter
-    - Following are available: (uppercase, snakecase, titlecase, kebabcase, camelcase, pascalcase, forwardslash, backslash, doublequote, indented, indented_8, indented_12, indented_16, escapeDollar)
-    - And our custom ones: (oneOfName, capitalised, unquote, jsname)
diff --git a/openai-client/generator/build.gradle.kts b/openai-client/generator/build.gradle.kts
deleted file mode 100644
index 8e3398049..000000000
--- a/openai-client/generator/build.gradle.kts
+++ /dev/null
@@ -1,42 +0,0 @@
-@file:Suppress("DSL_SCOPE_VIOLATION")
-
-plugins {
-    java
-    alias(libs.plugins.spotless)
-}
-
-dependencies {
-    implementation("org.openapitools:openapi-generator-cli:7.4.0")
-}
-
-tasks.test {
-    useJUnitPlatform()
-}
-
-task("downloadOpenAIAPI", JavaExec::class) {
-    group = "GenerateTasks"
-    mainClass = "ai.xef.openai.generator.DownloadOpenAIAPI"
-    classpath = sourceSets["main"].runtimeClasspath
-}
-
-task("openaiClientGenerate", JavaExec::class) {
-    group = "GenerateTasks"
-    mainClass = "org.openapitools.codegen.OpenAPIGenerator"
-    val command = (listOf(
-        "generate",
-        "-i",
-        "config/openai-api.yaml",
-        "-g",
-        "ai.xef.openai.generator.KMMGeneratorConfig",
-        "-o",
-        "../client/build/generated/OpenAI/",
-        "--skip-validate-spec"
-    ) + if (project.hasProperty("debugModels") && !project.hasProperty("debugOperations")) {
-        listOf("--global-property", "debugModels=true")
-    } else if(!project.hasProperty("debugModels") && project.hasProperty("debugOperations")) {
-        listOf("--global-property", "debugOperations=true")
-    } else emptyList())
-
-    args = command
-    classpath = sourceSets["main"].runtimeClasspath
-}
diff --git a/openai-client/generator/config/api.mustache b/openai-client/generator/config/api.mustache
deleted file mode 100644
index 0c18d55ef..000000000
--- a/openai-client/generator/config/api.mustache
+++ /dev/null
@@ -1,194 +0,0 @@
-{{>licenseInfo}}
-package {{apiPackage}}
-
-{{#imports}}import {{import}}
-{{/imports}}
-import com.xebia.functional.openai.Config
-import com.xebia.functional.openai.UploadFile
-import com.xebia.functional.openai.appendGen
-import com.xebia.functional.openai.generated.api.{{classname}}.*
-import com.xebia.functional.openai.streamEvents
-import com.xebia.functional.openai.errors.serializeOrThrowWithResponseInfo
-import io.ktor.client.HttpClient
-import io.ktor.client.call.body
-import io.ktor.client.plugins.timeout
-import io.ktor.client.request.HttpRequestBuilder
-import io.ktor.client.request.accept
-import io.ktor.client.request.header
-import io.ktor.client.request.forms.formData
-import io.ktor.client.request.parameter
-import io.ktor.client.request.prepareRequest
-import io.ktor.client.request.request
-import io.ktor.client.request.setBody
-import io.ktor.client.statement.HttpResponse
-import io.ktor.client.statement.HttpStatement
-import io.ktor.http.ContentType
-import io.ktor.http.HttpHeaders
-import io.ktor.http.HttpMethod
-import io.ktor.http.contentType
-import io.ktor.http.path
-import kotlinx.coroutines.flow.Flow
-import kotlinx.coroutines.flow.flow
-import kotlinx.serialization.*
-import kotlinx.serialization.descriptors.*
-import kotlinx.serialization.encoding.*
-import kotlinx.serialization.json.Json
-import kotlinx.serialization.json.JsonObject
-import kotlinx.serialization.json.jsonObject
-import kotlinx.serialization.json.JsonPrimitive
-import kotlin.time.Duration.Companion.seconds
-import kotlin.time.DurationUnit
-
-{{#operations}}
-/**
-  *
-  */
-interface {{classname}} {
-    {{#operation}}
-        {{#allParams}}
-            {{#isEnum}}
-
-                /**
-                * enum for parameter {{paramName}}
-                */
-                @Serializable
-                {{#nonPublicApi}}internal {{/nonPublicApi}}enum class {{enumName}}{{operationIdCamelCase}}(val value: {{^isContainer}}{{dataType}}{{/isContainer}}{{#isContainer}}kotlin.String{{/isContainer}}) {
-                {{^enumUnknownDefaultCase}}
-                    {{#allowableValues}}{{#enumVars}}
-                        @SerialName(value = {{^isString}}"{{/isString}}{{{value}}}{{^isString}}"{{/isString}})
-                        {{&name}}({{{value}}}){{^-last}},{{/-last}}
-                    {{/enumVars}}{{/allowableValues}}
-                {{/enumUnknownDefaultCase}}
-                {{#enumUnknownDefaultCase}}
-                    {{#allowableValues}}{{#enumVars}}{{^-last}}
-                        @SerialName(value = {{^isString}}"{{/isString}}{{{value}}}{{^isString}}"{{/isString}})
-                        {{&name}}({{{value}}}),
-                    {{/-last}}{{/enumVars}}{{/allowableValues}}
-                {{/enumUnknownDefaultCase}}
-                }
-
-            {{/isEnum}}
-        {{/allParams}}
-
-    /**
-    * {{summary}}
-    * {{notes}}
-    {{#allParams}} * @param {{{paramName}}} {{description}}{{^required}} (optional{{#defaultValue}}, default to {{{.}}}{{/defaultValue}}){{/required}}{{#required}}{{#defaultValue}} (default to {{{.}}}){{/defaultValue}}{{/required}}
-    {{/allParams}} * @param configure optional configuration for the request, allows overriding the default configuration.
-    * @return {{{returnType}}}{{^returnType}}void{{/returnType}}
-    */
-    suspend fun {{operationId}}({{#allParams}}{{{paramName}}}: {{#isEnum}}{{# isContainer}}kotlin.collections.List<{{enumName}}{{operationIdCamelCase}}>{{/isContainer}}{{^isContainer}}{{enumName}}{{operationIdCamelCase}}{{/isContainer}}{{/isEnum}}{{^isEnum}}{{{dataType}}}{{/isEnum}}{{#required}}{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{/required}}{{^required}}?{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{^defaultValue}} = null{{/defaultValue}}{{/required}}, {{/allParams}}configure: HttpRequestBuilder.() -> Unit = {}): {{#returnProperty}}{{#isFile}}HttpResponse{{/isFile}}{{^isFile}}{{{returnType}}}{{^returnType}}Unit{{/returnType}}{{/isFile}}{{/returnProperty}}
-
-    {{#vendorExtensions.x-streaming}}
-    /**
-    * Streaming variant: {{summary}}
-    * By default, the client is modified to timeout after 60 seconds. Which is overridable by the [configure].
-    * {{notes}}
-    {{#allParams}} * @param {{{paramName}}} {{description}}{{^required}} (optional{{#defaultValue}}, default to {{{.}}}{{/defaultValue}}){{/required}}{{#required}}{{#defaultValue}} (default to {{{.}}}){{/defaultValue}}{{/required}}
-    {{/allParams}} * @param configure optional configuration for the request, allows overriding the default configuration.
-    * @return [Flow]<[{{{vendorExtensions.x-streaming-return}}}]>
-    */
-    fun {{operationId}}Stream({{#allParams}}{{{paramName}}}: {{#isEnum}}{{# isContainer}}kotlin.collections.List<{{enumName}}{{operationIdCamelCase}}>{{/isContainer}}{{^isContainer}}{{enumName}}{{operationIdCamelCase}}{{/isContainer}}{{/isEnum}}{{^isEnum}}{{{dataType}}}{{/isEnum}}{{#required}}{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{/required}}{{^required}}?{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{^defaultValue}} = null{{/defaultValue}}{{/required}}, {{/allParams}}configure: HttpRequestBuilder.() -> Unit = {}): Flow<{{{vendorExtensions.x-streaming-return}}}>
-    {{/vendorExtensions.x-streaming}}
-    {{/operation}}
-}
-{{/operations}}
-
-{{#operations}}
-fun {{classname}}(client: HttpClient, config: Config): {{apiPackage}}.{{classname}} = object : {{apiPackage}}.{{classname}} {
-    {{#operation}}
-    override suspend fun {{operationId}}({{#allParams}}{{{paramName}}}: {{#isEnum}}{{# isContainer}}kotlin.collections.List<{{enumName}}{{operationIdCamelCase}}>{{/isContainer}}{{^isContainer}}{{enumName}}{{operationIdCamelCase}}{{/isContainer}}{{/isEnum}}{{^isEnum}}{{{dataType}}}{{/isEnum}}{{^required}}?{{/required}},{{/allParams}} configure: HttpRequestBuilder.() -> Unit): {{#returnProperty}}{{#isFile}}HttpResponse{{/isFile}}{{^isFile}}{{{returnType}}}{{^returnType}}Unit{{/returnType}}{{/isFile}}{{/returnProperty}} =
-      client.request {
-        configure()
-        method = HttpMethod.{{#lambda.capitalised}}{{httpMethod}}{{/lambda.capitalised}}
-        {{#headerParams}}
-            header("{{baseName}}", {{#isContainer}}this.joinToString(separator = collectionDelimiter("{{collectionFormat}}")){{/isContainer}}{{^isContainer}}this.toString(){{/isContainer}})
-        {{/headerParams}}
-        contentType(ContentType.Application.Json)
-        {{#queryParams}}
-            parameter("{{baseName}}", {{#isContainer}}toMultiValue(this, "{{collectionFormat}}"){{/isContainer}}{{^isContainer}}{{{paramName}}}{{/isContainer}})
-        {{/queryParams}}
-        url { path("{{#lambda.dropslash}}{{path}}{{/lambda.dropslash}}"{{#pathParams}}.replace("{" + "{{baseName}}" + "}", {{#isContainer}}{{paramName}}.joinToString(","){{/isContainer}}{{^isContainer}}"${{{paramName}}}"{{/isContainer}}){{/pathParams}}) }
-        setBody({{#hasBodyParam}}{{#bodyParam}}{{#isArray}}{{operationIdCamelCase}}Request({{{paramName}}}{{^isList}}.asList(){{/isList}}){{/isArray}}{{^isArray}}{{#isMap}}{{operationIdCamelCase}}Request({{{paramName}}}){{/isMap}}{{^isMap}}{{{paramName}}}{{/isMap}}{{/isArray}}{{/bodyParam}}{{/hasBodyParam}}
-        {{^hasBodyParam}}
-            {{#hasFormParams}}
-                {{#isMultipart}}
-                    formData {
-                    {{#formParams}}
-                        {{#isArray}}
-                            {{{paramName}}}?.onEach { appendGen("{{{baseName}}}[]", it) }
-                        {{/isArray}}
-                        {{^isArray}}
-                            {{{paramName}}}?.apply { appendGen("{{{baseName}}}", {{{paramName}}}) }
-                        {{/isArray}}
-                    {{/formParams}}
-                    }
-                {{/isMultipart}}
-                {{^isMultipart}}
-                    ParametersBuilder().also {
-                    {{#formParams}}
-                        {{{paramName}}}?.apply { it.append("{{{baseName}}}", {{{paramName}}}.toString()) }
-                    {{/formParams}}
-                    }.build()
-                {{/isMultipart}}
-            {{/hasFormParams}}
-            {{^hasFormParams}}
-                io.ktor.client.utils.EmptyContent
-            {{/hasFormParams}}
-        {{/hasBodyParam}})
-      }{{#returnProperty}}{{^isFile}}.serializeOrThrowWithResponseInfo(){{/isFile}}{{/returnProperty}}
-
-    {{#vendorExtensions.x-streaming}}
-    override fun {{operationId}}Stream({{#allParams}}{{{paramName}}}: {{#isEnum}}{{# isContainer}}kotlin.collections.List<{{enumName}}{{operationIdCamelCase}}>{{/isContainer}}{{^isContainer}}{{enumName}}{{operationIdCamelCase}}{{/isContainer}}{{/isEnum}}{{^isEnum}}{{{dataType}}}{{/isEnum}}{{#required}}{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{/required}}{{^required}}?{{#defaultValue}} = {{^isNumber}}{{#isEnum}}{{enumName}}{{operationIdCamelCase}}.{{enumDefaultValue}}{{/isEnum}}{{^isEnum}}{{{defaultValue}}}{{/isEnum}}{{/isNumber}}{{#isNumber}}{{{defaultValue}}}.toDouble(){{/isNumber}}{{/defaultValue}}{{^defaultValue}} = null{{/defaultValue}}{{/required}}, {{/allParams}}configure: HttpRequestBuilder.() -> Unit): Flow<{{{vendorExtensions.x-streaming-return}}}> = flow {
-      client.prepareRequest {
-        timeout {
-          requestTimeoutMillis = 60.seconds.toLong(DurationUnit.MILLISECONDS)
-          socketTimeoutMillis = 60.seconds.toLong(DurationUnit.MILLISECONDS)
-        }
-        configure()
-        method = HttpMethod.{{#lambda.capitalised}}{{httpMethod}}{{/lambda.capitalised}}
-        accept(ContentType.Text.EventStream)
-        header(HttpHeaders.CacheControl, "no-cache")
-        header(HttpHeaders.Connection, "keep-alive")
-        {{#headerParams}}
-        header("{{baseName}}", {{#isContainer}}this.joinToString(separator = collectionDelimiter("{{collectionFormat}}")){{/isContainer}}{{^isContainer}}this.toString(){{/isContainer}})
-        {{/headerParams}}
-        contentType(ContentType.Application.Json)
-        {{#queryParams}}
-            parameter("{{baseName}}", {{#isContainer}}toMultiValue(this, "{{collectionFormat}}"){{/isContainer}}{{^isContainer}}listOf("${{{paramName}}}"){{/isContainer}})
-        {{/queryParams}}
-        url { path("{{#lambda.dropslash}}{{path}}{{/lambda.dropslash}}"{{#pathParams}}.replace("{" + "{{baseName}}" + "}", {{#isContainer}}{{paramName}}.joinToString(","){{/isContainer}}{{^isContainer}}"${{{paramName}}}"{{/isContainer}}){{/pathParams}}) }
-        {{#hasBodyParam}}
-        val element = Json.encodeToJsonElement({{#lambda.serializer}}{{#bodyParam}}{{{baseName}}}{{/bodyParam}}{{/lambda.serializer}}, {{#bodyParam}}{{{paramName}}}{{/bodyParam}})
-        val jsObject = JsonObject(element.jsonObject + Pair("stream", JsonPrimitive(true)))
-        setBody(jsObject)
-        {{/hasBodyParam}}
-        {{^hasBodyParam}}
-            {{#hasFormParams}}
-                {{#isMultipart}}
-                    formData {
-                    {{#formParams}}
-                        {{#isArray}}
-                            {{{paramName}}}?.onEach { appendGen("{{{baseName}}}[]", it) }
-                        {{/isArray}}
-                        {{^isArray}}
-                            {{{paramName}}}?.apply { appendGen("{{{baseName}}}", {{{paramName}}}) }
-                        {{/isArray}}
-                    {{/formParams}}
-                    }
-                {{/isMultipart}}
-                {{^isMultipart}}
-                    ParametersBuilder().also {
-                    {{#formParams}}
-                        {{{paramName}}}?.apply { it.append("{{{baseName}}}", {{{paramName}}}.toString()) }
-                    {{/formParams}}
-                    }.build()
-                {{/isMultipart}}
-            {{/hasFormParams}}
-        {{/hasBodyParam}}
-      }.execute { streamEvents(it, config.json, config.streamingPrefix, config.streamingDelimiter) }
-    }
-    {{/vendorExtensions.x-streaming}}
-{{/operation}}
-}
-{{/operations}}
diff --git a/openai-client/generator/config/data_class.mustache b/openai-client/generator/config/data_class.mustache
deleted file mode 100644
index 6ccb7d95f..000000000
--- a/openai-client/generator/config/data_class.mustache
+++ /dev/null
@@ -1,68 +0,0 @@
-import kotlinx.serialization.Serializable
-import kotlinx.serialization.SerialName
-import kotlinx.serialization.Contextual
-import kotlin.js.JsName
-import kotlinx.serialization.json.*
-{{#enumUnknownDefaultCase}}
-import kotlinx.serialization.KSerializer
-import kotlinx.serialization.Serializer
-import kotlinx.serialization.builtins.serializer
-import kotlinx.serialization.encoding.Decoder
-import kotlinx.serialization.encoding.Encoder
-{{/enumUnknownDefaultCase}}
-{{#hasEnums}}
-{{/hasEnums}}
-
-/**
-* {{{description}}}
-*
-{{#allVars}}
-    * @param {{{name}}} {{{description}}}
-{{/allVars}}
-*/
-{{^discriminator}}@Serializable{{/discriminator}}
-{{#isDeprecated}}@Deprecated(message = "This schema is deprecated."){{/isDeprecated}}
-{{>additionalModelTypeAnnotations}}
-{{#nonPublicApi}}internal {{/nonPublicApi}}{{#discriminator}}interface{{/discriminator}}{{^discriminator}}data class{{/discriminator}} {{classname}}{{^discriminator}} (
-{{#allVars}}
-    {{#required}}{{>data_class_req_var}}{{/required}}{{^required}}{{>data_class_opt_var}}{{/required}}{{^-last}},{{/-last}}
-{{/allVars}}
-){{/discriminator}}{{#parent}} : {{{parent}}}{{#isMap}}(){{/isMap}}{{#isArray}}(){{/isArray}}{{/parent}} {
-{{#discriminator}}{{#vars}}{{#required}}
-    {{>interface_req_var}}{{/required}}{{^required}}
-    {{>interface_opt_var}}{{/required}}{{/vars}}{{/discriminator}}
-{{#hasEnums}}
-    {{#vars}}
-        {{#isEnum}}
-            /**
-            * {{{description}}}
-            *
-            * Values: {{#allowableValues}}{{#enumVars}}{{&name}}{{^-last}},{{/-last}}{{/enumVars}}{{/allowableValues}}
-            */
-            @Serializable{{#enumUnknownDefaultCase}}(with = {{classname}}Serializer::class){{/enumUnknownDefaultCase}}
-            {{#nonPublicApi}}internal {{/nonPublicApi}}enum class {{{nameInCamelCase}}}(val value: {{dataType}}) {
-            {{#allowableValues}}
-                {{#enumVars}}
-                @SerialName(value = {{#lambda.doublequote}}{{{value}}}{{/lambda.doublequote}}) {{#lambda.jsname}}{{&name}}{{/lambda.jsname}}({{{value}}}){{^-last}},{{/-last}}{{#-last}};{{/-last}}
-                {{/enumVars}}
-            {{/allowableValues}}
-                }{{#kotlinx_serialization}}{{#enumUnknownDefaultCase}}
-
-                    @Serializer(forClass = {{{nameInCamelCase}}}::class)
-                    internal object {{nameInCamelCase}}Serializer : KSerializer<{{nameInCamelCase}}> {
-                      override val descriptor = {{{dataType}}}.serializer().descriptor
-
-                      override fun deserialize(decoder: Decoder): {{nameInCamelCase}} {
-                        val value = decoder.decodeSerializableValue({{{dataType}}}.serializer())
-                        return {{nameInCamelCase}}.values().firstOrNull { it.value == value }
-                          ?: {{nameInCamelCase}}.{{#allowableValues}}{{#enumVars}}{{#-last}}{{&name}}{{/-last}}{{/enumVars}}{{/allowableValues}}
-                      }
-
-                      override fun serialize(encoder: Encoder, value: {{nameInCamelCase}}) {
-                        encoder.encodeSerializableValue({{{dataType}}}.serializer(), value.value)
-                      }
-                    }{{/enumUnknownDefaultCase}}{{/kotlinx_serialization}}
-        {{/isEnum}}
-    {{/vars}}
-{{/hasEnums}}
-}
\ No newline at end of file
diff --git a/openai-client/generator/config/data_class_opt_var.mustache b/openai-client/generator/config/data_class_opt_var.mustache
deleted file mode 100644
index d59b528d1..000000000
--- a/openai-client/generator/config/data_class_opt_var.mustache
+++ /dev/null
@@ -1,7 +0,0 @@
-{{#description}}
-    /* {{{.}}} */
-{{/description}}
-    {{#deprecated}}
-    @Deprecated(message = "This property is deprecated.")
-    {{/deprecated}}
-    @SerialName(value = "{{{vendorExtensions.x-base-name-literal}}}") {{#isInherited}}override {{/isInherited}}{{>modelMutable}} {{{name}}}: {{#isArray}}{{#isList}}{{#uniqueItems}}kotlin.collections.{{#modelMutable}}Mutable{{/modelMutable}}Set{{/uniqueItems}}{{^uniqueItems}}kotlin.collections.{{#modelMutable}}Mutable{{/modelMutable}}List{{/uniqueItems}}{{/isList}}{{^isList}}kotlin.Array{{/isList}}<{{^items.isEnum}}{{^items.isPrimitiveType}}{{^items.isModel}}{{#kotlinx_serialization}}@Contextual {{/kotlinx_serialization}}{{/items.isModel}}{{/items.isPrimitiveType}}{{{items.dataType}}}{{/items.isEnum}}{{#items.isEnum}}{{classname}}.{{{nameInCamelCase}}}{{/items.isEnum}}>{{/isArray}}{{^isEnum}}{{^isArray}}{{{dataType}}}{{/isArray}}{{/isEnum}}{{#isEnum}}{{^isArray}}{{classname}}.{{{nameInCamelCase}}}{{/isArray}}{{/isEnum}}? = {{^defaultValue}}null{{/defaultValue}}{{#defaultValue}}{{^isNumber}}{{{defaultValue}}}{{/isNumber}}{{#isNumber}}({{{defaultValue}}}).toDouble(){{/isNumber}}{{/defaultValue}}
\ No newline at end of file
diff --git a/openai-client/generator/config/data_class_req_var.mustache b/openai-client/generator/config/data_class_req_var.mustache
deleted file mode 100644
index 17b800584..000000000
--- a/openai-client/generator/config/data_class_req_var.mustache
+++ /dev/null
@@ -1,7 +0,0 @@
-{{#description}}
-    /* {{{.}}} */
-{{/description}}
-    {{#deprecated}}
-    @Deprecated(message = "This property is deprecated.")
-    {{/deprecated}}
-    @SerialName(value = "{{{vendorExtensions.x-base-name-literal}}}") {{#isInherited}}override {{/isInherited}}val {{{name}}}: {{#isArray}}{{#isList}}{{#uniqueItems}}kotlin.collections.Set{{/uniqueItems}}{{^uniqueItems}}kotlin.collections.List{{/uniqueItems}}{{/isList}}{{^isList}}kotlin.Array{{/isList}}<{{^items.isEnum}}{{^items.isPrimitiveType}}{{^items.isModel}}@Contextual {{/items.isModel}}{{/items.isPrimitiveType}}{{{items.dataType}}}{{/items.isEnum}}{{#items.isEnum}}{{classname}}.{{{nameInCamelCase}}}{{/items.isEnum}}>{{/isArray}}{{^isEnum}}{{^isArray}}{{{dataType}}}{{/isArray}}{{/isEnum}}{{#isEnum}}{{^isArray}}{{classname}}.{{{nameInCamelCase}}}{{/isArray}}{{/isEnum}}{{#isNullable}}?{{/isNullable}}{{#defaultValue}} = {{^isNumber}}{{{defaultValue}}}{{/isNumber}}{{#isNumber}}({{{defaultValue}}}).toDouble(){{/isNumber}}{{/defaultValue}}
\ No newline at end of file
diff --git a/openai-client/generator/config/enum_class.mustache b/openai-client/generator/config/enum_class.mustache
deleted file mode 100644
index 535b97aa5..000000000
--- a/openai-client/generator/config/enum_class.mustache
+++ /dev/null
@@ -1,53 +0,0 @@
-import kotlinx.serialization.*
-import kotlinx.serialization.builtins.*
-import kotlinx.serialization.encoding.*
-import kotlin.jvm.JvmStatic
-{{#allowableValues}}{{#enumVars}}import com.xebia.functional.openai.generated.model.{{classname}}.Supported.{{#lambda.unquote}}{{&value}}{{/lambda.unquote}}
-{{/enumVars}}{{/allowableValues}}
-
-/**
- * {{{description}}}
- */
-// We define a serializer for the parent sum type, and then use it to serialize the child types
-@Suppress("SERIALIZER_TYPE_INCOMPATIBLE")
-@Serializable(with = {{classname}}Serializer::class)
-sealed interface {{classname}} {
-  val value: {{{dataType}}}
-
-  @Serializable(with = {{classname}}Serializer::class)
-  enum class Supported(override val value: {{{dataType}}}) : {{classname}} {
-    {{#allowableValues}}{{#enumVars}}@SerialName(value = {{{value}}}) {{#lambda.unquote}}{{&value}}{{/lambda.unquote}}({{#isArray}}{{#isList}}listOf({{{value}}})){{/isList}}{{^isList}}arrayOf({{{value}}})){{/isList}}{{/isArray}}{{^isArray}}{{#isString}}{{#lambda.doublequote}}{{{value}}}{{/lambda.doublequote}}{{/isString}}{{^isString}}{{{value}}}{{/isString}}{{/isArray}}){{^-last}},{{/-last}}{{#-last}};{{/-last}}{{/enumVars}}{{/allowableValues}}
-
-    override fun toString(): kotlin.String = value{{^isString}}.toString(){{/isString}}
-  }
-
-  @Serializable(with = {{classname}}Serializer::class)
-  data class Custom(override val value: kotlin.String) : {{classname}}
-
-    companion object {
-      @JvmStatic
-      fun valueOf(value: {{{dataType}}}): {{classname}} =
-        values().firstOrNull { it.value == value } ?: Custom(value)
-
-      {{#allowableValues}}{{#enumVars}}
-          inline val {{#lambda.unquote}}{{&value}}{{/lambda.unquote}}: {{classname}}
-            get() = Supported.{{#lambda.unquote}}{{&value}}{{/lambda.unquote}}
-      {{/enumVars}}{{/allowableValues}}
-
-      @JvmStatic fun values(): List<{{classname}}> = Supported.entries
-    }
-}
-
-object {{classname}}Serializer : KSerializer<{{classname}}> {
-    private val valueSerializer = {{{dataType}}}.serializer()
-    override val descriptor = valueSerializer.descriptor
-
-    override fun deserialize(decoder: Decoder): {{classname}} {
-        val value = decoder.decodeSerializableValue(valueSerializer)
-        return {{classname}}.valueOf(value)
-    }
-
-    override fun serialize(encoder: Encoder, value: {{classname}}) {
-        encoder.encodeSerializableValue(valueSerializer, value.value)
-    }
-}
diff --git a/openai-client/generator/config/model.mustache b/openai-client/generator/config/model.mustache
deleted file mode 100644
index bf6667742..000000000
--- a/openai-client/generator/config/model.mustache
+++ /dev/null
@@ -1,17 +0,0 @@
-{{>licenseInfo}}
-package {{modelPackage}}
-
-{{#imports}}import {{import}}
-{{/imports}}
-{{#models}}
-{{#model}}
-{{! We execute this block only for the _first_ element: if oneOf is not empty, execute the template }}
-{{#oneOf}}{{#-first}}{{>model_oneof}}{{/-first}}{{/oneOf}}{{^oneOf}}
-{{#isEnum}}{{>enum_class}}{{/isEnum}}{{^isEnum}}
-{{#isAlias}}typealias {{classname}} = {{{dataType}}}{{/isAlias}}{{^isAlias}}
-{{>data_class}}
-{{/isAlias}}
-{{/isEnum}}
-{{/oneOf}}
-{{/model}}
-{{/models}}
\ No newline at end of file
diff --git a/openai-client/generator/config/model_oneof.mustache b/openai-client/generator/config/model_oneof.mustache
deleted file mode 100644
index 132176c5d..000000000
--- a/openai-client/generator/config/model_oneof.mustache
+++ /dev/null
@@ -1,41 +0,0 @@
-import com.xebia.functional.openai.attemptDeserialize
-import kotlin.jvm.JvmInline
-import kotlinx.serialization.*
-import kotlinx.serialization.builtins.*
-import kotlinx.serialization.descriptors.*
-import kotlinx.serialization.encoding.*
-import kotlinx.serialization.json.*
-
-@Serializable(with = {{classname}}Serializer::class)
-sealed interface {{classname}} {
-
-  {{#oneOf}}
-  @JvmInline
-  @Serializable
-  value class Case{{#lambda.oneOfName}}{{{.}}}{{/lambda.oneOfName}}(val value: {{{.}}}): {{classname}}
-  {{/oneOf}}
-
-}
-
-object {{classname}}Serializer : KSerializer<{{classname}}> {
-    @OptIn(InternalSerializationApi::class, ExperimentalSerializationApi::class)
-    override val descriptor: SerialDescriptor = buildSerialDescriptor("{{classname}}", PolymorphicKind.SEALED) { {{#oneOf}}
-      element("{{#lambda.oneOfName}}{{-index}}{{/lambda.oneOfName}}", {{#lambda.serializer}}{{{.}}}{{/lambda.serializer}}.descriptor){{/oneOf}}
-    }
-
-    override fun deserialize(decoder: Decoder): {{classname}} {
-      val jsonSerde = Json { ignoreUnknownKeys = true }
-      val json = decoder.decodeSerializableValue(JsonElement.serializer())
-      return attemptDeserialize(json, {{#oneOf}}
-        Pair({{classname}}.Case{{#lambda.oneOfName}}{{{.}}}{{/lambda.oneOfName}}::class) {
-          {{classname}}.Case{{#lambda.oneOfName}}{{{.}}}{{/lambda.oneOfName}}(jsonSerde.decodeFromJsonElement({{#lambda.serializer}}{{{.}}}{{/lambda.serializer}}, json))
-      }{{#-last}}){{/-last}}{{^-last}},{{/-last}}{{/oneOf}}
-    }
-
-    override fun serialize(encoder: Encoder, value: {{classname}}) =
-      when(value) { {{#oneOf}}
-        is {{classname}}.Case{{#lambda.oneOfName}}{{{.}}}{{/lambda.oneOfName}} ->
-        encoder.encodeSerializableValue({{#lambda.serializer}}{{{.}}}{{/lambda.serializer}}, value.value)
-        {{/oneOf}}
-    }
-}
diff --git a/openai-client/generator/config/openai-api-commit b/openai-client/generator/config/openai-api-commit
deleted file mode 100644
index 94b46cf74..000000000
--- a/openai-client/generator/config/openai-api-commit
+++ /dev/null
@@ -1 +0,0 @@
-df5699f5c48fe455d9d036198e77fdb637edee29
diff --git a/openai-client/generator/config/openai.mustache b/openai-client/generator/config/openai.mustache
deleted file mode 100644
index f9cc7a5fc..000000000
--- a/openai-client/generator/config/openai.mustache
+++ /dev/null
@@ -1,26 +0,0 @@
-{{#apiInfo}}
-{{>licenseInfo}}
-package {{apiPackage}}
-
-import com.xebia.functional.openai.Config
-import io.ktor.client.HttpClient
-{{#imports}}import {{import}}
-{{/imports}}
-{{#apis}}import com.xebia.functional.openai.generated.api.{{classFilename}}
-{{/apis}}
-
-interface OpenAI : AutoCloseable {
-{{#apis}}
-  val {{#lambda.camelcase}}{{classFilename}}{{/lambda.camelcase}}: {{classFilename}}
-{{/apis}}
-}
-
-fun OpenAI(client: HttpClient, config: Config): OpenAI = object : OpenAI, AutoCloseable {
-{{#apis}}
-  override val {{#lambda.camelcase}}{{classFilename}}{{/lambda.camelcase}} = {{classFilename}}(client, config)
-{{/apis}}
-
-  override fun close() = client.close()
-}
-
-{{/apiInfo}}
\ No newline at end of file
diff --git a/openai-client/generator/gradle.properties b/openai-client/generator/gradle.properties
deleted file mode 100644
index 1eda1733c..000000000
--- a/openai-client/generator/gradle.properties
+++ /dev/null
@@ -1 +0,0 @@
-org.gradle.jvmargs="-Dlog.level=debug --global-property debugModels"
\ No newline at end of file
diff --git a/openai-client/generator/src/main/java/ai/xef/openai/generator/DownloadOpenAIAPI.java b/openai-client/generator/src/main/java/ai/xef/openai/generator/DownloadOpenAIAPI.java
deleted file mode 100644
index e0570e9fc..000000000
--- a/openai-client/generator/src/main/java/ai/xef/openai/generator/DownloadOpenAIAPI.java
+++ /dev/null
@@ -1,40 +0,0 @@
-package ai.xef.openai.generator;
-
-import java.io.*;
-import java.net.URL;
-import java.nio.channels.Channels;
-import java.nio.channels.FileChannel;
-import java.nio.channels.ReadableByteChannel;
-
-public class DownloadOpenAIAPI {
-    public static void main(String[] args) {
-        try {
-            String commit = readCommit();
-            downloadAPI(commit);
-
-            // list streaming methods
-        } catch (IOException e) {
-            throw new RuntimeException(e);
-        }
-    }
-
-    private static String readCommit() throws IOException {
-        StringBuilder resultStringBuilder = new StringBuilder();
-        try (BufferedReader br = new BufferedReader(new FileReader("config/openai-api-commit"))) {
-            String line;
-            while ((line = br.readLine()) != null) {
-                resultStringBuilder.append(line).append("\n");
-            }
-        }
-        return resultStringBuilder.toString().trim();
-    }
-
-    private static void downloadAPI(String commit) throws IOException {
-        URL url = new URL("https://raw.githubusercontent.com/openai/openai-openapi/%s/openapi.yaml".formatted(commit));
-        ReadableByteChannel readableByteChannel = Channels.newChannel(url.openStream());
-        try (FileOutputStream fileOutputStream = new FileOutputStream("config/openai-api.yaml")) {
-            FileChannel fileChannel = fileOutputStream.getChannel();
-            fileChannel.transferFrom(readableByteChannel, 0, Long.MAX_VALUE);
-        }
-    }
-}
diff --git a/openai-client/generator/src/main/java/ai/xef/openai/generator/KMMGeneratorConfig.java b/openai-client/generator/src/main/java/ai/xef/openai/generator/KMMGeneratorConfig.java
deleted file mode 100644
index 2017ec89c..000000000
--- a/openai-client/generator/src/main/java/ai/xef/openai/generator/KMMGeneratorConfig.java
+++ /dev/null
@@ -1,308 +0,0 @@
-package ai.xef.openai.generator;
-
-import com.google.common.collect.ImmutableMap;
-import com.samskivert.mustache.Mustache;
-import com.samskivert.mustache.Template;
-import org.apache.commons.lang3.NotImplementedException;
-import org.apache.commons.lang3.tuple.Pair;
-import org.openapitools.codegen.*;
-import org.openapitools.codegen.languages.KotlinClientCodegen;
-import org.openapitools.codegen.model.ModelMap;
-import org.openapitools.codegen.model.ModelsMap;
-import org.openapitools.codegen.model.OperationsMap;
-
-import java.io.IOException;
-import java.io.Writer;
-import java.util.*;
-
-import static java.util.Map.entry;
-
-@SuppressWarnings("unused")
-public class KMMGeneratorConfig extends KotlinClientCodegen {
-
-
-    private final Map<String, List<String>> nonRequiredFields = new LinkedHashMap<>();
-
-    @Override
-    public String getIgnoreFilePathOverride() {
-        return ".openapi-generator-ignore";
-    }
-
-    public KMMGeneratorConfig() {
-        super();
-
-        // Generate in src/commonMain/kotlin, not /src/main/kotlin
-        additionalProperties.put("sourceFolder", "src/commonMain/kotlin");
-        additionalProperties.put("generateModelTests", false);
-        additionalProperties.put("generateApiTests", false);
-        additionalProperties.put("generateInfrastructure", false);
-        setModelPackage("com.xebia.functional.openai.generated.model");
-        setApiPackage("com.xebia.functional.openai.generated.api");
-        additionalProperties.put("apiSuffix", "");
-        additionalProperties.put("modelSuffix", "");
-
-        // Configure OpenAI `object` to be mapped to `JsonObject`
-        typeMapping.put("object", "JsonObject");
-        importMapping.put("JsonObject", "kotlinx.serialization.json.JsonObject");
-
-        typeMapping.put("java.net.URI", "kotlin.String");
-
-        typeMapping.put("java.math.BigDecimal", "kotlin.Double");
-        importMapping.put("BigDecimal", "kotlin.Double");
-
-        typeMapping.put("java.io.File", "UploadFile");
-        importMapping.put("java.io.File", "com.xebia.functional.openai.UploadFile");
-
-        // Maps `Map<String, Any>` to `JsonObject`
-        schemaMapping.put("FunctionParameters", "kotlinx.serialization.json.JsonObject");
-
-        // Configure the template directory
-        templateDir = "config";
-        supportingFiles.add(new SupportingFile("openai.mustache", "src/commonMain/kotlin/" + apiPackage.replace(".", "/"), "OpenAI.kt"));
-
-        omitGradleWrapper = true;
-        serializationLibrary = SERIALIZATION_LIBRARY_TYPE.kotlinx_serialization;
-
-        defaultIncludes.remove("io.ktor.client.request.forms.InputProvider");
-        defaultIncludes.add("com.xebia.functional.openai.apis.UploadFile");
-
-        importMapping.remove("InputProvider");
-        importMapping.put("UploadFile", "com.xebia.functional.openai.apis.UploadFile");
-
-        // Fixes for DateTime
-        dateLibrary = DateLibrary.KOTLINX_DATETIME.value;
-        typeMapping.put("date", "kotlinx.datetime.LocalDate");
-        typeMapping.put("date-time", "kotlinx.datetime.Instant");
-        typeMapping.put("DateTime", "Instant");
-        importMapping.put("Instant", "kotlinx.datetime.Instant");
-
-        specialCharReplacements.put("-", "_");
-        specialCharReplacements.put(".", "_");
-        // TODO PascalCase!?!
-        enumPropertyNaming = CodegenConstants.ENUM_PROPERTY_NAMING_TYPE.snake_case;
-        nonRequiredFields.putAll(
-                Map.ofEntries(
-                        entry("ListAssistantFilesResponse", List.of("firstId", "lastId")),
-                        entry("ListAssistantsResponse", List.of("firstId", "lastId")),
-                        entry("ListMessageFilesResponse", List.of("firstId", "lastId")),
-                        entry("ListMessagesResponse", List.of("firstId", "lastId")),
-                        entry("ListRunsResponse", List.of("firstId", "lastId")),
-                        entry("ListRunStepsResponse", List.of("firstId", "lastId")),
-                        entry("ListThreadsResponse", List.of("firstId", "lastId")),
-                        entry("MessageObject", List.of("status", "metadata")),
-                        entry("MessageObjectContentInner", List.of("imageFile", "text")),
-                        entry("RunObject", List.of("expiresAt", "requiredAction")),
-                        entry("RunStepDetailsToolCallsCodeObjectCodeInterpreterOutputsInner", List.of("logs", "image")),
-                        entry("RunStepDetailsToolCallsFunctionObjectFunction", List.of("name", "arguments", "output")),
-                        entry("RunStepDetailsToolCallsFunctionObject", List.of("id")),
-                        entry("RunStepDetailsToolCallsObjectToolCallsInner", List.of("codeInterpreter", "retrieval", "function")),
-                        entry("RunStepDetailsToolCallsRetrievalObject", List.of("retrieval")),
-                        entry("RunStepObject", List.of("expiredAt", "metadata")),
-                        entry("MessageContentTextObjectTextAnnotationsInner", List.of("filePath", "fileCitation"))
-                )
-        );
-    }
-
-    private Optional<CodegenProperty> readEnumModel(List<CodegenProperty> all) {
-        if (all.size() == 2) {
-            CodegenProperty first = all.get(0);
-            CodegenProperty second = all.get(1);
-            if (first.isString && second.isEnum) {
-                return Optional.of(second);
-            }
-        }
-        return Optional.empty();
-    }
-
-    /**
-     * Map<OperationId, StreamedReturnType>
-     * Used to generate additional code for operations that support streaming.
-     * <p>
-     * Extra streaming operation will be generated for OperationId, and the return type will be Flow<StreamedReturnType>.
-     */
-    private final static Map<String, Pair<String, String>> streamingOps = Map.of(
-            "createThreadAndRun", Pair.of("com.xebia.functional.openai", "ServerSentEvent"),
-            "createRun", Pair.of("com.xebia.functional.openai", "ServerSentEvent"),
-            "submitToolOuputsToRun", Pair.of("com.xebia.functional.openai", "ServerSentEvent"),
-            "createChatCompletion", Pair.of("com.xebia.functional.openai.generated.model", "CreateChatCompletionStreamResponse")
-    );
-
-    /**
-     * Add the `x-streaming` vendor extension to the operations that are streaming,
-     * and add `x-streaming-return` of the return type of the operation.
-     * <p>
-     * This is used in the mustache template to generate additional code for operations that support streaming.
-     */
-    @Override
-    public OperationsMap postProcessOperationsWithModels(OperationsMap objs, List<ModelMap> allModels) {
-        objs.getOperations()
-                .getOperation()
-                .forEach((op) -> {
-                            if (streamingOps.containsKey(op.operationId)) {
-                                op.vendorExtensions.put("x-streaming", true);
-                                Pair<String, String> returnType = streamingOps.get(op.operationId);
-                                Map<String, String> imports = Map.of(
-                                        "import", returnType.getKey() + "." + returnType.getValue(),
-                                        "classname", returnType.getValue()
-                                );
-                                if(!objs.getImports().contains(imports)) {
-                                    objs.getImports().add(imports);
-                                }
-                                op.vendorExtensions.put("x-streaming-return", returnType.getValue());
-                            }
-                        }
-                );
-        return super.postProcessOperationsWithModels(objs, allModels);
-    }
-
-    @Override
-    public ModelsMap postProcessModels(ModelsMap objs) {
-        for (ModelMap mo : objs.getModels()) {
-            CodegenModel cm = mo.getModel();
-            if (cm.anyOf != null && !cm.anyOf.isEmpty()) {
-                Optional<CodegenProperty> codegenProperty = readEnumModel(cm.getComposedSchemas().getAnyOf());
-                codegenProperty.ifPresent((enumProp) -> {
-                    cm.modelJson = enumProp.jsonSchema;
-                    cm.interfaces = null;
-                    cm.anyOf = new HashSet<>();
-                    cm.dataType = enumProp.dataType;
-                    cm.isString = enumProp.isString;
-                    cm.allowableValues = enumProp.allowableValues;
-                    cm.isEnum = enumProp.isEnum;
-                    cm.setIsAnyType(enumProp.isAnyType);
-                    cm.setComposedSchemas(null);
-                });
-            } else if (nonRequiredFields.containsKey(cm.classname)) {
-                List<String> fields = nonRequiredFields.getOrDefault(cm.classname, Collections.emptyList());
-                cm
-                        .allVars
-                        .stream()
-                        .filter(p -> fields.contains(p.name))
-                        .forEach(p -> p.setRequired(false));
-            }
-        }
-        return super.postProcessModels(objs);
-    }
-
-    @Override
-    protected ImmutableMap.Builder<String, Mustache.Lambda> addMustacheLambdas() {
-        return super.addMustacheLambdas()
-                .put("oneOfName", new OneOfName())
-                .put("capitalised", new Capitalised())
-                .put("unquote", new Unquote())
-                .put("jsname", new JsName())
-                .put("serializer", new Serializer())
-                .put("dropslash", new DropSlash());
-    }
-
-    /* Mechanism to do array access in mustache...
-     * We need to generate names for the cases of `oneOf`,
-     * where we generate a `sealed interface` with the Schema name,
-     * and a `data class CaseInnerType(val value: InnerType)` for each of the `oneOf` cases. */
-    public static class OneOfName implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            serializer(writer, fragment.execute().trim(), 0);
-        }
-
-        private void serializer(
-                Writer buffer,
-                String text,
-                int depth
-        ) throws IOException {
-            if (text.startsWith("kotlin.collections.List<")) {
-                String inner = text.substring(24, text.length() - 1);
-                serializer(buffer, inner, depth + 1);
-            } else if (text.startsWith("kotlin.collections.")) {
-                throw new NotImplementedException(text + " collection serialization not supported **yet**");
-            } else {
-                if (text.startsWith("kotlin.")) {
-                    text = text.substring(7);
-                }
-                buffer.write(text);
-                if (depth > 0) {
-                    buffer.write("s");
-                    depth--;
-                }
-                while (depth-- > 0) {
-                    buffer.write("List");
-                }
-            }
-        }
-    }
-
-    /* Lambda to generate the `@JsName` annotation for the `length` property,
-     * can be generalised to other properties/names if needed */
-    public static class JsName implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            String text = fragment.execute();
-            if (text.equals("length")) {
-                writer.write("@JsName(\"length_type\") length");
-            } else {
-                writer.write(text);
-            }
-        }
-    }
-
-    /* Lambda to capitalise the first letter of a string, and lowercase the rest */
-    public static class Capitalised implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            String text = fragment.execute();
-            writer.write(text.substring(0, 1).toUpperCase() + text.substring(1).toLowerCase(Locale.ROOT));
-        }
-    }
-
-    /* Drop first `/` from a path */
-    public static class DropSlash implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            String text = fragment.execute();
-            if (text.startsWith("/")) {
-                writer.write(text.substring(1));
-            } else {
-                writer.write(text);
-            }
-        }
-    }
-
-    /* enum with `gpt-` are trimmed by `gpt_`, the raw value needs to trim the surrounding `"` */
-    public static class Unquote implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            String text = snakeCase(fragment.execute());
-            if (text.startsWith("\"") && text.endsWith("\"")) {
-                writer.write(text.substring(1, text.length() - 1));
-            } else {
-                writer.write(text);
-            }
-        }
-
-        // Poor man lower snakeCase
-        private String snakeCase(String text) {
-            return text.replace("-", "_").replace(".", "_");
-        }
-    }
-
-    /* Most advanced lambda, to generate the `serializer()` call for the `kotlinx.serialization` library.
-     * This works for ListSerializer, but can work for much more!! */
-    public static class Serializer implements Mustache.Lambda {
-        public void execute(Template.Fragment fragment, Writer writer) throws IOException {
-            serializer(writer, fragment.execute().trim(), 0);
-        }
-
-        private void serializer(
-                Writer buffer,
-                String text,
-                int depth
-        ) throws IOException {
-            if (text.startsWith("kotlin.collections.List<")) {
-                String inner = text.substring(24, text.length() - 1);
-                buffer.write("ListSerializer(");
-                serializer(buffer, inner, depth + 1);
-            } else if (text.startsWith("Map<")) {
-                throw new NotImplementedException("Map serialization not supported **yet**");
-            } else {
-                buffer.write(text);
-                buffer.write(".serializer()");
-                buffer.write(")".repeat(Math.max(0, depth)));
-            }
-        }
-    }
-}
diff --git a/server/src/main/kotlin/com/xebia/functional/xef/server/services/PostgresVectorStoreService.kt b/server/src/main/kotlin/com/xebia/functional/xef/server/services/PostgresVectorStoreService.kt
index 6f764ea4e..6b8f74d81 100644
--- a/server/src/main/kotlin/com/xebia/functional/xef/server/services/PostgresVectorStoreService.kt
+++ b/server/src/main/kotlin/com/xebia/functional/xef/server/services/PostgresVectorStoreService.kt
@@ -1,8 +1,8 @@
 package com.xebia.functional.xef.server.services
 
-import com.xebia.functional.openai.generated.model.CreateEmbeddingRequestModel
 import com.xebia.functional.xef.Config
 import com.xebia.functional.xef.OpenAI
+import com.xebia.functional.xef.openapi.CreateEmbeddingRequest
 import com.xebia.functional.xef.store.PGVectorStore
 import com.xebia.functional.xef.store.VectorStore
 import com.xebia.functional.xef.store.postgresql.PGDistanceStrategy
@@ -21,8 +21,8 @@ class PostgresVectorStoreService(
   private val preDeleteCollection: Boolean = false,
   private val chunkSize: Int = 400,
   private val distanceStrategy: PGDistanceStrategy = PGDistanceStrategy.Euclidean,
-  private val embeddingRequestModel: CreateEmbeddingRequestModel =
-    CreateEmbeddingRequestModel.text_embedding_3_small
+  private val embeddingRequestModel: CreateEmbeddingRequest.Model =
+    CreateEmbeddingRequest.Model.TextEmbedding3Small
 ) : VectorStoreService() {
 
   fun addCollection() {
diff --git a/settings.gradle.kts b/settings.gradle.kts
index dbb1da79e..ebf2bdfed 100644
--- a/settings.gradle.kts
+++ b/settings.gradle.kts
@@ -26,13 +26,6 @@ project(":xef-filesystem").projectDir = file("filesystem")
 
 include("xef-tokenizer")
 project(":xef-tokenizer").projectDir = file("tokenizer")
-
-include("xef-openai-client")
-project(":xef-openai-client").projectDir = file("openai-client/client")
-
-include("xef-openai-client-generator")
-project(":xef-openai-client-generator").projectDir = file("openai-client/generator")
-
 //</editor-fold>
 
 //<editor-fold desc="Integrations">