> ## Documentation Index
> Fetch the complete documentation index at: https://platform.atlan.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To act on Atlan objects, use the Atlan MCP server at https://api.atlan.com/mcp or the SDK. To install or publish skills from a terminal, use the `aireg` CLI; `aireg --map json` prints its command map. The `atlanai` CLI is deprecated. Run a read-only identity check before any write.
> The docs MCP server at /mcp searches these docs only. It cannot read or change Atlan objects.
> SDK packages: Python `atlanai` (PyPI) and TypeScript `@atlanai/sdk` (npm). Show Python first, then TypeScript.
> These docs cover Atlan AI Registry, Atlan's hosted product. The AI registry specification and aireg CLI are maintained separately. /open-source lists publicly reachable signed installers and explains that the source repository and hosted CLI reference currently require GitHub access. Do not claim the repository or its reference docs are publicly accessible. Use these docs for Atlan-specific setup such as the adapter, sign-in and workspaces.

# Model Gateway

> Call every model your organization connects through one OpenAI- and Anthropic-compatible endpoint, with provider keys kept in Atlan.

Model Gateway gives your applications and coding agents one endpoint for every model your organization connects. You keep your OpenAI or Anthropic SDK, use one Atlan key, and see what each call cost.

```bash theme={null}
curl https://api.atlan.com/openai/v1/chat/completions \
  -H "Authorization: Bearer $ATLAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "<provider>/<model>", "messages": [{"role": "user", "content": "Hello"}]}'
```

```mermaid theme={null}
flowchart TD
  A["Your app or agent"] -->|"Atlan API key"| G["Model Gateway"]
  G -->|"provider credential"| P["Connected provider"]
  G --> U["Usage and cost"]
```

## What you can do

<CardGroup cols={2}>
  <Card title="Call any connected model" icon="plug" href="/gateway/model/how-tos/connect-a-client">
    Point an OpenAI or Anthropic SDK at Atlan and call it.
  </Card>

  <Card title="Move existing code over" icon="right-left" href="/gateway/model/how-tos/migrate-existing-calls">
    Change a base URL, key, and model name. Nothing else.
  </Card>

  <Card title="Connect your providers" icon="server" href="/gateway/model/how-tos/connect-providers">
    Add provider credentials once and choose which models agents call.
  </Card>

  <Card title="Fall back across providers" icon="shuffle" href="/gateway/model/how-tos/routing-and-fallback">
    Spread calls over deployments, retry, and keep them in-region.
  </Card>

  <Card title="Reuse identical answers" icon="bolt" href="/gateway/model/how-tos/cache-responses">
    Serve a repeated request from cache without calling the provider.
  </Card>

  <Card title="See who spends what" icon="chart-line" href="/gateway/model/how-tos/review-usage">
    Review usage by model, person, and agent, down to each call.
  </Card>
</CardGroup>

## Get started

<Card title="Connect client to Model Gateway" icon="rocket" href="/gateway/model/how-tos/connect-a-client">
  Create an agent key, list the models you can call, and make your first request.
</Card>

## Key concepts

| Term | Meaning |
| - | - |
| [Provider](/gateway/model/concepts/how-it-works#providers-and-model-names) | A connected account, such as your OpenAI organization or an AWS Bedrock region. |
| [Model name](/gateway/model/concepts/how-it-works#providers-and-model-names) | What you send as `model`: `<provider>/<model>`, copied from `GET /openai/v1/models`. |
| [Dialect](/gateway/model/concepts/how-it-works#translation-between-dialects) | The API shape you call in: OpenAI at `/openai/v1` or Anthropic at `/anthropic/v1`. |
| [Modality](/gateway/model/concepts/how-it-works#modalities) | What a model is for, such as `chat` or `embedding`. It decides which endpoint serves it. |
| [Route and fall back across deployments](/gateway/model/how-tos/routing-and-fallback) | Several models behind one name, chosen by weight, with fallback between them. |

## Common tasks

| I want to… | Go to |
| - | - |
| Make my first call | [Connect client to Model Gateway](/gateway/model/how-tos/connect-a-client) |
| Move an app off a direct OpenAI or Anthropic key | [Migrate existing model calls](/gateway/model/how-tos/migrate-existing-calls) |
| Understand what happens to a request | [How Model Gateway works](/gateway/model/concepts/how-it-works) |
| Add a provider or turn a model on or off | [Connect model providers](/gateway/model/how-tos/connect-providers) |
| Set up fallback or region limits for a model | [Route and fall back across deployments](/gateway/model/how-tos/routing-and-fallback) |
| Cache repeated requests | [Cache identical responses](/gateway/model/how-tos/cache-responses) |
| Find a call I just made, or see spend by person | [Review model usage](/gateway/model/how-tos/review-usage) |
| Check whether a field or endpoint is supported | [Model API compatibility](/gateway/model/references/compatibility) |

## Availability and limits

* Only organization admins add providers and turn models on or off. Everyone else calls the models that are on.
* A model with no price lands off, and can't be turned on until it has rates.
* The gateway serves selected OpenAI and Anthropic operations. Files, batches, assistants, vector stores, fine-tuning, and Anthropic token counting return `404`.
* The Responses API is stateless: each request carries its whole conversation.
* Calls an agent makes with its own key appear in usage reports for organization admins only.

## Reference

* [Model API compatibility](/gateway/model/references/compatibility): endpoints, headers, model names, and request limits.
* [Supported model providers](/gateway/model/references/providers): provider types, credentials, and discovery.
* [List available models](/api-reference/models/list-available-models): the qualified names your client can call.
* [OpenAI-compatible chat completion](/api-reference/models/openai-compatible-chat-completion): request and response schemas.

## Troubleshooting

See [Model Gateway errors](/gateway/model/troubleshooting) for refused calls and models missing from `GET /models`.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.