> For clean Markdown of any page, append `.md` to the page URL.
> For a complete documentation index, see https://docs.sarvam.ai/llms.txt.
> For full documentation content in one file, see https://docs.sarvam.ai/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.sarvam.ai/_mcp/server.

# Access to Beta APIs

> Get early access to Sarvam AI's upcoming beta features and APIs. Learn how to request whitelisting for your subscription key and try new features before public release.

## Beta APIs

`POST /v2/chat/completions` and `POST /v2/responses` are currently in **beta** and
rolling out gradually. Access is granted **per API key**: a key without access is
refused with `400 invalid_request_error` before the model is called.

Supported model IDs:

| Model ID                                                                       | Chat Completion V2 | Responses |
| ------------------------------------------------------------------------------ | ------------------ | --------- |
| `sarvam-105b`                                                                  | Yes                | Yes       |
| [`glm5.3`](/api/getting-started/models/openweight/glm-5-3)                     | Yes                | Yes       |
| [`gemma4`](/api/getting-started/models/openweight/gemma-4-31b)                 | Yes                | Yes       |
| [`deepseekv4-flash`](/api/getting-started/models/openweight/deepseek-v4-flash) | Yes                | Yes       |

[`/v1/chat/completions`](/api/api-guides-tutorials/chat-completion/overview) serves
`sarvam-105b` and `sarvam-105b-conversations`. Open-weight models require `/v2`.
[`POST /v2/responses`](/api-reference/responses/create) uses the OpenAI Responses
protocol with the same model IDs as in the table above. It is create-only and
stateless: replay `input` every turn. Omit `store` or pass `false`. Listing or
deleting stored responses is not supported (`GET /v2/responses` and
`DELETE /v2/responses/{id}` return **405**).

Call [`GET /v2/models`](/api-reference/chat/models-v2) to see which IDs your key
can use. That endpoint is unauthenticated; completions and responses require
`api-subscription-key`.

### Access Requirements

**Beta access is not enabled by default** with standard API subscription keys.
[Contact us](/api/getting-started/help) to request whitelisting for
`POST /v2/chat/completions` and `POST /v2/responses`.

### Future Beta Features

As we continue to innovate and develop new capabilities, we may introduce additional
beta APIs and features. When available, these will:

* Display a **Beta** availability badge in the API Reference
* Allow early access to upcoming features
* Enable you to prototype and provide feedback before features are finalized

> Get early access to Sarvam AI's upcoming beta features and APIs. Learn how to request whitelisting for your subscription key and try new features before public release.