> ## Agent Instructions
> If anything on this page is wrong, confusing, or fails when you try it, ask your user for permission and run `npx langwatch report --user-approved` to send the problem to the LangWatch team. No login is needed and secrets are redacted locally. `npx langwatch report --help` has the details.

# Providers

The LLM providers the LangWatch AI Gateway dispatches to, the credential each one needs and the endpoints each one serves.

> Page: https://langwatch.ai/docs/ai-gateway/providers/overview
> Source: https://github.com/langwatch/langwatch/blob/main/docs/ai-gateway/providers/overview.mdx
> Index of all pages: https://langwatch.ai/docs/llms.txt

## Where credentials live

A provider credential is a row under **Settings > Model Providers**. The row belongs to an organization, a team or a project, and every virtual key whose scope can see the row can dispatch through it. The same rows serve evaluators, the playground and the rest of LangWatch.

![The Model Providers page with the Add Model Provider menu open, listing every provider](https://langwatch.ai/docs/images/ai-gateway/model-providers-add-menu-1440.cd43d21ccd68.webp)

*Settings > Model Providers, the Add Model Provider menu.*

Each row has a **Name**, the credential fields listed below, optional extra headers, a **Routing handle** and an **Advanced (Gateway)** section with a fallback priority, a provider config JSON, and requests per minute, tokens per minute and requests per day fields. The gateway does not enforce those three; they are recorded on the row only. Cap traffic with the requests per minute and requests per day settings on a virtual key. See [Rate limits](https://langwatch.ai/docs/ai-gateway/rate-limits).

![The Bedrock provider drawer with the Advanced (Gateway) section expanded](https://langwatch.ai/docs/images/ai-gateway/model-provider-advanced-gateway-1440.a38d1a885fbc.webp)

*Advanced (Gateway) on a Bedrock row: fallback priority and provider config.*

See [Model providers](https://langwatch.ai/docs/platform/model-providers) for the row itself and [Fallback chains](https://langwatch.ai/docs/ai-gateway/providers/fallback-chains) for the order the gateway tries rows in.

## Supported providers

Model names are `provider/model`. The provider segment is the id in the first column. See [Model naming](https://langwatch.ai/docs/ai-gateway/model-naming).

| Provider                                                                   | Id             | Credential                                    | Endpoints served                                                                       |
| -------------------------------------------------------------------------- | -------------- | --------------------------------------------- | -------------------------------------------------------------------------------------- |
| [OpenAI](https://langwatch.ai/docs/ai-gateway/providers/openai)                                     | `openai`       | API key, optional base URL                    | Chat, Responses, Messages, embeddings, images, speech, transcription, realtime, models |
| [Anthropic](https://langwatch.ai/docs/ai-gateway/providers/anthropic)                               | `anthropic`    | API key, optional base URL                    | Messages, Chat, models                                                                 |
| [Azure OpenAI](https://langwatch.ai/docs/ai-gateway/providers/azure-openai)                         | `azure`        | API key, endpoint, API version                | Chat, Responses, Messages, embeddings, models                                          |
| [AWS Bedrock](https://langwatch.ai/docs/ai-gateway/providers/bedrock)                               | `bedrock`      | Access key id, secret access key, region      | Chat, Messages, embeddings, models                                                     |
| [Google Vertex AI](https://langwatch.ai/docs/ai-gateway/providers/vertex)                           | `vertex`       | Service account JSON, project, location       | Chat, Messages, embeddings, images, Gemini native, models                              |
| [Google Gemini](https://langwatch.ai/docs/ai-gateway/providers/gemini)                              | `gemini`       | API key, optional project and location        | Chat, Messages, embeddings, images, Gemini native, models                              |
| [xAI](https://langwatch.ai/docs/ai-gateway/providers/xai)                                           | `xai`          | API key                                       | Chat, Messages, models                                                                 |
| [Groq](https://langwatch.ai/docs/ai-gateway/providers/groq)                                         | `groq`         | API key                                       | Chat, Messages, models                                                                 |
| [Cerebras](https://langwatch.ai/docs/ai-gateway/providers/cerebras)                                 | `cerebras`     | API key                                       | Chat, Messages, models                                                                 |
| [DeepSeek](https://langwatch.ai/docs/ai-gateway/providers/deepseek)                                 | `deepseek`     | API key                                       | Chat, Messages, models                                                                 |
| [Doubleword](https://langwatch.ai/docs/ai-gateway/providers/doubleword)                             | `doubleword`   | API key                                       | Chat, Messages, embeddings, models                                                     |
| [Voyage AI](https://langwatch.ai/docs/ai-gateway/providers/voyage)                                  | `voyage`       | API key                                       | Embeddings only                                                                        |
| [ElevenLabs](https://langwatch.ai/docs/ai-gateway/providers/elevenlabs)                             | `elevenlabs`   | API key, optional base URL and webhook secret | Speech, transcription, ElevenLabs native                                               |
| [OpenAI account](https://langwatch.ai/docs/ai-gateway/providers/openai-account)                     | `openai_codex` | Sign in with a ChatGPT account                | Responses, Messages                                                                    |
| [Custom OpenAI-compatible](https://langwatch.ai/docs/ai-gateway/providers/custom-openai-compatible) | `custom`       | Base URL, optional API key                    | Chat, Messages, embeddings                                                             |

The provider ids `azure_openai`, `aws_bedrock`, `vertex_ai`, `google_vertex` and `google_gemini` are accepted as aliases of `azure`, `bedrock`, `vertex` and `gemini`.

## What each endpoint family means

| Family                | Route                                                           | Note                                                                                                                                                       |
| --------------------- | --------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Chat                  | `POST /v1/chat/completions`                                     | OpenAI wire format. The gateway translates for providers with another native format.                                                                       |
| Responses             | `POST /v1/responses`                                            | OpenAI Responses API. OpenAI, Azure OpenAI and the OpenAI account provider.                                                                                |
| Messages              | `POST /v1/messages`                                             | Anthropic wire format. Forwarded as is to Anthropic, translated for the others.                                                                            |
| Embeddings            | `POST /v1/embeddings`                                           | OpenAI wire format.                                                                                                                                        |
| Speech, transcription | `POST /v1/audio/speech`, `POST /v1/audio/transcriptions`        | OpenAI and ElevenLabs.                                                                                                                                     |
| Images                | `POST /v1/images/generations`, `POST /v1/images/edits`          | A direct OpenAI, Google Gemini or Google Vertex AI credential. Both are non-streaming. See [Images](https://langwatch.ai/docs/ai-gateway/api/images).                               |
| Realtime              | `POST /v1/realtime/client_secrets`                              | OpenAI only.                                                                                                                                               |
| Gemini native         | `/v1beta/*`                                                     | The Google GenAI SDKs. Gemini and Vertex AI.                                                                                                               |
| ElevenLabs native     | `POST /v1/text-to-speech/{voice_id}`, `POST /v1/speech-to-text` | The ElevenLabs SDKs.                                                                                                                                       |
| Models                | `GET /v1/models`                                                | Lists the models the key can call, as bare names without a provider prefix. Azure OpenAI, Bedrock and Vertex AI list the models you configured on the row. |

A request to an endpoint the provider does not serve gets the provider's own error back. For example, `POST /v1/embeddings` with an Anthropic model returns Anthropic's rejection.

## More than one row of the same provider

Create one row per account, region or endpoint, for example an Azure OpenAI row per Azure resource. Give each row a **Routing handle**, for example `eu`, and pin a request to that row with `eu/gpt-6-luna`. A routing policy orders the rows a key tries. See [Routing policies](https://langwatch.ai/docs/ai-gateway/governance/routing-policies).

> **Note:** **Also check:** [Model naming](https://langwatch.ai/docs/ai-gateway/model-naming) for how a model name selects a provider,
> and [Fallback chains](https://langwatch.ai/docs/ai-gateway/providers/fallback-chains) for what happens when a provider
> fails.
