diff --git a/content/docs/ai-gateway/gemini.md b/content/docs/ai-gateway/gemini.md index 80b293e0d9..b350a56170 100644 --- a/content/docs/ai-gateway/gemini.md +++ b/content/docs/ai-gateway/gemini.md @@ -6,7 +6,7 @@ summary: >- streamGenerateContent APIs through Neon AI Gateway. Use the google-genai SDK with a custom base URL. enableTableOfContents: true -updatedOn: '2026-11-18T00:17:54.883Z' +updatedOn: '2026-11-20T16:51:06.128Z' --- The Gemini endpoint exposes the [Google Gemini API](https://ai.google.dev/api/generate-content) through Neon AI Gateway. Use it when you're already working with the `google-genai` SDK and want to keep your existing code. For most use cases, the [chat completions](/docs/ai-gateway/chat-completions) endpoint is simpler to set up and works with Gemini models via the OpenAI SDK. @@ -34,18 +34,20 @@ NEON_AI_GATEWAY_BASE_URL=https://br-winter-pond-aptw82ef-api.ai.c-2.us-east-2.aw This endpoint accepts Google models only: -| Model ID | Notes | -| ----------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | -| `gemini-4-carbon` | Supports `temperature` and `thinkingConfig.thinkingLevel` on this endpoint, but not on [chat completions](/docs/ai-gateway/chat-completions#switching-models). | -| `gemini-4-argon` | Supports `temperature` and `thinkingConfig.thinkingLevel` on this endpoint, but not on [chat completions](/docs/ai-gateway/chat-completions#switching-models). | -| `gemini-3-8-flash` | | -| `gemini-3-7-flash` | | -| `gemini-3-6-flash` | | -| `gemini-3-5-flash` | | -| `gemini-3-5-flash-lite` | | -| `gemini-3-1-pro` | | -| `gemini-3-1-flash-lite` | | -| `gemini-3-flash` | | +| Model ID | Notes | +| ------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| `gemini-4-carbon` | Supports `temperature` and `thinkingConfig.thinkingLevel` on this endpoint, but not on [chat completions](/docs/ai-gateway/chat-completions#switching-models). | +| `gemini-4-argon` | Supports `temperature` and `thinkingConfig.thinkingLevel` on this endpoint, but not on [chat completions](/docs/ai-gateway/chat-completions#switching-models). | +| `gemini-3-pro-image` | Generates images. On [chat completions](/docs/ai-gateway/chat-completions), the image comes back as an `image_url` content block. | +| `gemini-3-1-flash-image` | Generates images. On [chat completions](/docs/ai-gateway/chat-completions), the image comes back as an `image_url` content block. | +| `gemini-3-8-flash` | | +| `gemini-3-7-flash` | | +| `gemini-3-6-flash` | | +| `gemini-3-5-flash` | | +| `gemini-3-5-flash-lite` | | +| `gemini-3-1-pro` | | +| `gemini-3-1-flash-lite` | | +| `gemini-3-flash` | | Sending a non-Google model ID returns `400 model "" is not available on the gemini_generate_content endpoint`, naming whichever model you sent. Use the [chat completions endpoint](/docs/ai-gateway/chat-completions) if you want to call Gemini models alongside other providers from the same code. diff --git a/content/docs/ai-gateway/models.md b/content/docs/ai-gateway/models.md index 99daea920d..b009daa6e5 100644 --- a/content/docs/ai-gateway/models.md +++ b/content/docs/ai-gateway/models.md @@ -6,7 +6,7 @@ summary: >- models, which endpoint to use, shorter and longer request paths, the /v1/models API, rate limits, pricing, and provider terms. enableTableOfContents: true -updatedOn: '2026-11-18T00:17:54.883Z' +updatedOn: '2026-11-20T16:51:06.128Z' --- This page covers how to call models through Neon AI Gateway and the rules that apply. For the full list of models you can call, see the [model catalog](https://neon.com/ai-gateway#models). @@ -86,6 +86,7 @@ All paths below are appended to your branch's bare AI Gateway host (`NEON_AI_GAT On `/v1/chat/completions`, `message.content` is a plain string for most models. For the models below, the gateway returns an array of typed content blocks instead. This is a known gateway issue that we're fixing. - `gemini-3-7-flash`, `gemini-3-8-flash`, `gemini-4-argon`, and `gemini-4-carbon` return an array on every non-streaming response, even for a trivial prompt. Their streamed `delta.content` values are strings, with `null` on the final chunk. +- `gemini-3-pro-image` and `gemini-3-1-flash-image` return an array on every non-streaming response, and some streamed `delta.content` values are arrays too. When the model generates an image, the array includes an `image_url` block with the image as a base64 `data:image/png` URL. - `claude-opus-5-5`, `claude-haiku-5-5`, `claude-fable-5-1`, and `claude-fable-5-2` return an array whenever the model reasons and a string when it doesn't. They decide on their own when to reason, so even a trivial prompt can return an array. When they reason while streaming, some `delta.content` values are arrays too. A low `max_tokens` value can cut a response off before the `text` block appears, leaving only a `reasoning` block. The arrays have these shapes: @@ -96,6 +97,12 @@ The arrays have these shapes: { type: 'text', text: '...', thoughtSignature: '...' }, ]; +// Gemini image models, when they generate an image +[ + { type: 'reasoning', summary: [...] }, + { type: 'image_url', image_url: { url: 'data:image/png;base64,...' } }, +]; + // Claude models, once they reason [ { type: 'reasoning', summary: [...] }, diff --git a/content/docs/ai-gateway/openai-responses.md b/content/docs/ai-gateway/openai-responses.md index c3774cdd92..a3f0fe5ed3 100644 --- a/content/docs/ai-gateway/openai-responses.md +++ b/content/docs/ai-gateway/openai-responses.md @@ -6,7 +6,7 @@ summary: >- AI Gateway. Required for codex model variants, which do not work with the chat completions endpoint. enableTableOfContents: true -updatedOn: '2026-11-10T18:56:17.167Z' +updatedOn: '2026-11-20T16:51:06.128Z' --- The OpenAI Responses endpoint exposes the [OpenAI Responses API](https://platform.openai.com/docs/api-reference/responses) through Neon AI Gateway. Use it with the OpenAI SDK's `responses.create()` method by changing only the `baseURL`. @@ -37,6 +37,7 @@ This endpoint accepts OpenAI models only: | Model ID | Notes | | --------------- | ---------------------- | | `gpt-6-astra` | | +| `gpt-6-1-terra` | | | `gpt-6-1-sol` | | | `gpt-6-sol` | | | `gpt-6-luna` | | @@ -173,7 +174,7 @@ const result = streamText({ return result.toUIMessageStreamResponse(); ``` -AI SDK `generateImage()` is not supported by AI Gateway; image generation is available only through this Responses tool. +AI SDK `generateImage()` is not supported by AI Gateway. For OpenAI image models, use this Responses tool. The Google image models, `gemini-3-pro-image` and `gemini-3-1-flash-image`, generate images on [chat completions](/docs/ai-gateway/chat-completions) and the [Gemini endpoint](/docs/ai-gateway/gemini) instead. ## Error handling diff --git a/src/app/models.json/data.json b/src/app/models.json/data.json index db9fd859ca..50b6a38ed4 100644 --- a/src/app/models.json/data.json +++ b/src/app/models.json/data.json @@ -2139,6 +2139,62 @@ } } }, + "gpt-6-1-terra": { + "id": "gpt-6-1-terra", + "name": "GPT-6.1 Terra", + "lab": "openai", + "provider": "databricks", + "family": "gpt-terra", + "released": true, + "attachment": true, + "reasoning": true, + "reasoning_options": [ + { + "type": "effort", + "values": ["none", "low", "medium", "high", "xhigh"] + } + ], + "tool_call": true, + "temperature": false, + "structured_output": true, + "open_weights": false, + "knowledge": "2026-06-01", + "release_date": "2026-10-27", + "last_updated": "2026-10-27", + "modalities": { + "input": ["text", "image", "pdf"], + "output": ["text"] + }, + "limit": { + "context": 1050000, + "input": 922000, + "output": 128000 + }, + "cost": { + "input": 1, + "output": 5, + "cache_read": 0.05, + "cache_write": 1.25, + "tiers": [ + { + "input": 2, + "output": 7.5, + "cache_read": 0.1, + "cache_write": 2.5, + "tier": { + "type": "context", + "size": 272000 + } + } + ], + "context_over_200k": { + "input": 2, + "output": 7.5, + "cache_read": 0.1, + "cache_write": 2.5 + } + } + }, "gpt-6-luna": { "id": "gpt-6-luna", "name": "GPT-6 Luna", @@ -2414,9 +2470,9 @@ "output": 65536 }, "cost": { - "input": 0, - "output": 0, - "cache_read": 0 + "input": 2, + "output": 10, + "cache_read": 0.2 } }, "claude-fable-5-2": { @@ -2486,9 +2542,82 @@ "output": 65536 }, "cost": { - "input": 0, - "output": 0, - "cache_read": 0 + "input": 4, + "output": 20, + "cache_read": 0.2 + } + }, + "gemini-3-pro-image": { + "id": "gemini-3-pro-image", + "name": "Gemini 3 Pro Image", + "lab": "google", + "provider": "databricks", + "family": "gemini-pro", + "released": true, + "attachment": true, + "reasoning": true, + "reasoning_options": [ + { + "type": "toggle" + }, + { + "type": "effort", + "values": ["low", "medium", "high", "xhigh", "max"] + } + ], + "tool_call": true, + "temperature": true, + "structured_output": true, + "open_weights": false, + "knowledge": "2025-01", + "release_date": "2025-11-20", + "last_updated": "2025-11-20", + "modalities": { + "input": ["text", "image", "pdf"], + "output": ["text", "image"] + }, + "limit": { + "context": 65536, + "output": 32768 + }, + "cost": { + "input": 2, + "output": 12 + } + }, + "gemini-3-1-flash-image": { + "id": "gemini-3-1-flash-image", + "name": "Gemini 3.1 Flash Image", + "lab": "google", + "provider": "databricks", + "family": "gemini-flash", + "released": true, + "attachment": true, + "reasoning": true, + "reasoning_options": [ + { + "type": "effort", + "values": ["low", "medium", "high", "max"] + } + ], + "tool_call": true, + "temperature": true, + "structured_output": true, + "open_weights": false, + "knowledge": "2025-01", + "release_date": "2026-02-26", + "last_updated": "2026-02-26", + "modalities": { + "input": ["text", "image", "pdf"], + "output": ["text", "image"] + }, + "limit": { + "context": 131072, + "output": 32768 + }, + "cost": { + "input": 0.5, + "output": 3 } } } diff --git a/src/app/models/capabilities.json b/src/app/models/capabilities.json index 2744dc1026..bbe0515cd8 100644 --- a/src/app/models/capabilities.json +++ b/src/app/models/capabilities.json @@ -748,6 +748,49 @@ "responses": false, "webSearch": false, "imageGeneration": false + }, + { + "id": "gpt-6-1-terra", + "owner": "openai", + "entitled": true, + "chat": "conforms", + "deviations": [], + "nativeDialect": "openai-responses", + "responses": true, + "webSearch": true, + "imageGeneration": true + }, + { + "id": "gemini-3-pro-image", + "owner": "google", + "entitled": true, + "chat": "array-content", + "deviations": [ + "`id` is null; the spec requires a string", + "`content` is an array[2] of {reasoning, text} with keys {summary, text, thoughtSignature, type}; the spec requires a string", + "`usage.total_tokens` is 964, but prompt + completion is 225", + "`usage.reasoning_tokens` is top-level; the spec nests it under `completion_tokens_details`" + ], + "nativeDialect": "gemini", + "responses": false, + "webSearch": false, + "imageGeneration": false + }, + { + "id": "gemini-3-1-flash-image", + "owner": "google", + "entitled": true, + "chat": "array-content", + "deviations": [ + "`id` is null; the spec requires a string", + "`content` is an array[1] of {text} with keys {text, thoughtSignature, type}; the spec requires a string", + "`usage.total_tokens` is 788, but prompt + completion is 242", + "`usage.reasoning_tokens` is top-level; the spec nests it under `completion_tokens_details`" + ], + "nativeDialect": "gemini", + "responses": false, + "webSearch": false, + "imageGeneration": false } ] }