Integrate SCX.ai as a provider for GLM-5.2 and Qwen3.8 Max

packages/__tests__/cost/__snapshots__/registrySnapshots.test.ts.snap

@@ -299,6 +299,30 @@ exports[`Registry Snapshots endpoint configurations snapshot 1`] = `
       ],
     },
   },
+  "alibaba/qwen3.8": {
+    "qwen3.8-max:scx": {
+      "context": 1000000,
+      "crossRegion": false,
+      "maxTokens": 131072,
+      "modelId": "Qwen3.8-Max",
+      "parameters": [
+        "frequency_penalty",
+        "max_tokens",
+        "presence_penalty",
+        "reasoning",
+        "response_format",
+        "stop",
+        "temperature",
+        "tools",
+        "top_p",
+      ],
+      "provider": "scx",
+      "ptbEnabled": true,
+      "regions": [
+        "*",
+      ],
+    },
+  },
   "anthropic/claude-3-haiku-20240307": {
     "claude-3-haiku-20240307:anthropic": {
       "context": 200000,
@@ -7292,6 +7316,34 @@ exports[`Registry Snapshots endpoint configurations snapshot 1`] = `
       ],
     },
   },
+  "zai/glm-5": {
+    "glm-5.2:scx": {
+      "context": 1000000,
+      "crossRegion": false,
+      "maxTokens": 131072,
+      "modelId": "GLM-5.2",
+      "parameters": [
+        "frequency_penalty",
+        "functions",
+        "max_tokens",
+        "presence_penalty",
+        "reasoning",
+        "response_format",
+        "seed",
+        "stop",
+        "structured_outputs",
+        "temperature",
+        "tool_choice",
+        "tools",
+        "top_p",
+      ],
+      "provider": "scx",
+      "ptbEnabled": true,
+      "regions": [
+        "*",
+      ],
+    },
+  },
 }
 `;
 
@@ -7312,6 +7364,9 @@ exports[`Registry Snapshots model coverage snapshot 1`] = `
     "novita",
     "openrouter",
   ],
+  "alibaba/qwen3.8": [
+    "scx",
+  ],
   "anthropic/claude-3-haiku-20240307": [
     "anthropic",
     "helicone",
@@ -7732,6 +7787,9 @@ exports[`Registry Snapshots model coverage snapshot 1`] = `
     "novita",
     "novita",
   ],
+  "zai/glm-5": [
+    "scx",
+  ],
 }
 `;
 
@@ -7790,6 +7848,18 @@ exports[`Registry Snapshots pricing snapshot 1`] = `
       },
     ],
   },
+  "alibaba/qwen3.8": {
+    "scx": [
+      {
+        "cacheMultipliers": {
+          "cachedInput": 0.1157,
+        },
+        "input": 0.000001815,
+        "output": 0.0000054461,
+        "threshold": 0,
+      },
+    ],
+  },
   "anthropic/claude-3-haiku-20240307": {
     "anthropic": [
       {
@@ -9953,6 +10023,18 @@ exports[`Registry Snapshots pricing snapshot 1`] = `
       },
     ],
   },
+  "zai/glm-5": {
+    "scx": [
+      {
+        "cacheMultipliers": {
+          "cachedInput": 0.25,
+        },
+        "input": 6.6e-7,
+        "output": 0.0000023106,
+        "threshold": 0,
+      },
+    ],
+  },
 }
 `;
 
@@ -10278,6 +10360,12 @@ exports[`Registry Snapshots verify registry state 1`] = `
         "novita",
       ],
     },
+    {
+      "model": "glm-5.2",
+      "providers": [
+        "scx",
+      ],
+    },
     {
       "model": "gpt-4.1",
       "providers": [
@@ -10813,6 +10901,12 @@ exports[`Registry Snapshots verify registry state 1`] = `
         "novita",
       ],
     },
+    {
+      "model": "qwen3.8-max",
+      "providers": [
+        "scx",
+      ],
+    },
     {
       "model": "sonar",
       "providers": [
@@ -10921,6 +11015,10 @@ exports[`Registry Snapshots verify registry state 1`] = `
       "modelCount": 5,
       "provider": "perplexity",
     },
+    {
+      "modelCount": 2,
+      "provider": "scx",
+    },
     {
       "modelCount": 23,
       "provider": "vertex",
@@ -10967,6 +11065,7 @@ exports[`Registry Snapshots verify registry state 1`] = `
     "gemma2-9b-it",
     "glm-4.6",
     "glm-4.7",
+    "glm-5.2",
     "gpt-4.1",
     "gpt-4.1-mini",
     "gpt-4.1-nano",
@@ -11034,6 +11133,7 @@ exports[`Registry Snapshots verify registry state 1`] = `
     "qwen3-coder-30b-a3b-instruct",
     "qwen3-next-80b-a3b-instruct",
     "qwen3-vl-235b-a22b-instruct",
+    "qwen3.8-max",
     "sonar",
     "sonar-deep-research",
     "sonar-pro",
@@ -11049,9 +11149,9 @@ exports[`Registry Snapshots verify registry state 1`] = `
     "claude-3.5-haiku:anthropic:*",
   ],
   "totalArchivedConfigs": 0,
-  "totalEndpoints": 329,
-  "totalModelProviderConfigs": 329,
-  "totalModelsWithPtb": 108,
-  "totalProviders": 21,
+  "totalEndpoints": 331,
+  "totalModelProviderConfigs": 331,
+  "totalModelsWithPtb": 110,
+  "totalProviders": 22,
 }
 `;

packages/cost/models/authors/alibaba/index.ts

@@ -8,19 +8,23 @@ import type { ModelConfig, ModelProviderConfig } from "../../types";
 // Import models
 import { models as qwen25Models } from "./qwen2.5/models";
 import { models as qwen3Models } from "./qwen3/models";
+import { models as qwen38Models } from "./qwen3.8/models";
 
 // Import endpoints
 import { endpoints as qwen25Endpoints } from "./qwen2.5/endpoints";
 import { endpoints as qwen3Endpoints } from "./qwen3/endpoints";
+import { endpoints as qwen38Endpoints } from "./qwen3.8/endpoints";
 
 // Aggregate models
 export const alibabaModels = {
   ...qwen25Models,
   ...qwen3Models,
+  ...qwen38Models,
 } satisfies Record<string, ModelConfig>;
 
 // Aggregate endpoints
 export const alibabaEndpointConfig = {
   ...qwen25Endpoints,
   ...qwen3Endpoints,
+  ...qwen38Endpoints,
 } satisfies Record<string, ModelProviderConfig>;

packages/cost/models/authors/alibaba/qwen3.8/endpoints.ts

@@ -0,0 +1,45 @@
+import { ModelProviderName } from "../../../providers";
+import type { ModelProviderConfig } from "../../../types";
+import { Qwen38ModelName } from "./models";
+
+export const endpoints = {
+  "qwen3.8-max:scx": {
+    providerModelId: "Qwen3.8-Max",
+    provider: "scx",
+    author: "alibaba",
+    pricing: [
+      {
+        threshold: 0,
+        input: 0.000001815,
+        output: 0.0000054461,
+        cacheMultipliers: {
+          cachedInput: 0.1157, // $0.21/M cache read vs $1.815/M input
+        },
+      },
+    ],
+    contextLength: 1_000_000,
+    maxCompletionTokens: 131_072,
+    // Qwen thinking models reject tool_choice "required" or an object, so
+    // tool_choice is intentionally left out.
+    supportedParameters: [
+      "max_tokens",
+      "temperature",
+      "top_p",
+      "stop",
+      "frequency_penalty",
+      "presence_penalty",
+      "reasoning",
+      "tools",
+      "response_format",
+    ],
+    ptbEnabled: true,
+    endpointConfigs: {
+      "*": {},
+    },
+  },
+} satisfies Partial<
+  Record<
+    `${Qwen38ModelName}:${ModelProviderName}` | Qwen38ModelName,
+    ModelProviderConfig
+  >
+>;

packages/cost/models/authors/alibaba/qwen3.8/models.ts

@@ -0,0 +1,17 @@
+import { ModelConfig } from "../../../types";
+
+export const models = {
+  "qwen3.8-max": {
+    name: "Qwen3.8 Max",
+    author: "alibaba",
+    description:
+      "Flagship 2.4-trillion-parameter MoE model in the Qwen3.8 series with native visual understanding, excelling at coding, professional work, and long-horizon autonomous agent tasks.",
+    contextLength: 1_000_000,
+    maxOutputTokens: 131_072,
+    created: "2026-08-02T00:00:00.000Z",
+    modality: { inputs: ["text", "image"], outputs: ["text"] },
+    tokenizer: "Qwen",
+  },
+} satisfies Record<string, ModelConfig>;
+
+export type Qwen38ModelName = keyof typeof models;

packages/cost/models/authors/zai/glm-5/endpoints.ts

@@ -0,0 +1,49 @@
+import { ModelProviderName } from "../../../providers";
+import type { ModelProviderConfig } from "../../../types";
+import { Glm5ModelName } from "./models";
+
+export const endpoints = {
+  "glm-5.2:scx": {
+    providerModelId: "GLM-5.2",
+    provider: "scx",
+    author: "zai",
+    pricing: [
+      {
+        threshold: 0,
+        input: 0.00000066,
+        output: 0.0000023106,
+        cacheMultipliers: {
+          cachedInput: 0.25, // $0.165/M cache read vs $0.66/M input
+        },
+      },
+    ],
+    quantization: "fp8",
+    contextLength: 1_000_000,
+    // SCX rejects max_tokens above 131072 even though the model card advertises 128k
+    maxCompletionTokens: 131_072,
+    supportedParameters: [
+      "functions",
+      "structured_outputs",
+      "reasoning",
+      "tool_choice",
+      "tools",
+      "response_format",
+      "max_tokens",
+      "temperature",
+      "top_p",
+      "stop",
+      "frequency_penalty",
+      "presence_penalty",
+      "seed",
+    ],
+    ptbEnabled: true,
+    endpointConfigs: {
+      "*": {},
+    },
+  },
+} satisfies Partial<
+  Record<
+    `${Glm5ModelName}:${ModelProviderName}` | Glm5ModelName,
+    ModelProviderConfig
+  >
+>;

packages/cost/models/authors/zai/glm-5/models.ts

@@ -0,0 +1,17 @@
+import type { ModelConfig } from "../../../types";
+
+export const models = {
+  "glm-5.2": {
+    name: "Zai GLM-5.2",
+    author: "zai",
+    description:
+      "GLM-5.2 is Zhipu AI's flagship model for long-horizon coding and agentic engineering tasks. It extends the GLM series to a 1M token context window and adds stronger multi-step reasoning and tool orchestration.",
+    contextLength: 1_000_000,
+    maxOutputTokens: 131_072,
+    created: "2026-06-13T00:00:00.000Z",
+    modality: { inputs: ["text"], outputs: ["text"] },
+    tokenizer: "Zai",
+  },
+} satisfies Record<string, ModelConfig>;
+
+export type Glm5ModelName = keyof typeof models;

packages/cost/models/authors/zai/index.ts

@@ -8,13 +8,17 @@ import type { ModelConfig, ModelProviderConfig } from "../../types";
 // Import models and endpoints
 import { models as glm4Models } from "./glm-4/models";
 import { endpoints as glm4Endpoints } from "./glm-4/endpoints";
+import { models as glm5Models } from "./glm-5/models";
+import { endpoints as glm5Endpoints } from "./glm-5/endpoints";
 
 // Aggregate models
 export const zaiModels = {
-  ...glm4Models
+  ...glm4Models,
+  ...glm5Models
 } satisfies Record<string, ModelConfig>;
 
 // Aggregate endpoints
 export const zaiEndpointConfig = {
-  ...glm4Endpoints
+  ...glm4Endpoints,
+  ...glm5Endpoints
 } satisfies Record<string, ModelProviderConfig>;

packages/cost/models/provider-helpers.ts

@@ -62,6 +62,8 @@ export function heliconeProviderToModelProviderName(
       return "fireworks";
     case "CANOPYWAVE":
       return "canopywave";
+    case "SCX":
+      return "scx";
     // new registry does not have
     case "LOCAL":
     case "HELICONE":
@@ -138,6 +140,9 @@ export const dbProviderToProvider = (
   if (provider === "canopywave" || provider === "Canopy Wave") {
     return "canopywave";
   }
+  if (provider === "scx" || provider === "SCX.ai") {
+    return "scx";
+  }
   if (provider === "novita" || provider === "Novita") {
     return "novita";
   }