> ## 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.

# Supported model providers

> Compare model-provider connections and their discovery and credential requirements.

Model Gateway supports the provider types below. Available models depend on the provider account, region, credentials, and the catalog your organization enables.

| Provider | Connection | Discovery and constraints |
| - | - | - |
| OpenAI | API key; optional approved endpoint override | Discovers the provider's model catalog. Supported operations depend on the model. |
| Anthropic | API key; optional approved endpoint override | Claude through the Messages API. Azure AI Foundry Claude uses the Anthropic provider type and its resource's `/anthropic/v1` URL; register deployments by name. |
| OpenRouter | API key | Discovers OpenRouter models. Upstream capability and pricing vary by model. |
| Amazon Bedrock | AWS region and an AWS credential, or deployment identity | Uses native Bedrock inference. Access depends on the region, model permissions, and any inference profile. |
| Google Vertex AI | Project and location URL; service account or deployment identity | Discovers Gemini models and uses native Gemini inference. Claude deployments use `anthropic/<model>@<version>` and are registered explicitly. |
| Google AI | API key | Uses Gemini's OpenAI-compatible endpoint. |
| Azure OpenAI / AI Foundry | Resource's OpenAI v1 endpoint and API key | Uses the resource endpoint and deployment names. Claude uses the Anthropic connection above. |
| OpenAI-compatible | Approved `/v1` base URL and API key | Uses the server's OpenAI-compatible contract; support varies by server and model. |
| LiteLLM proxy | Proxy root URL and API key | Discovery reads `/model/info`. Use a key that can list the intended models and configure proxy aliases for the upstream model IDs. |
| TypeSafe | API key | Decision models only. It does not serve chat, embedding, or media requests. |

<CardGroup cols={3}>
  <Card title="OpenAI" icon="https://mintcdn.com/atlan-602e2b74/pPcK1s6Qs7BTlBW2/assets/integrations/openai.svg?fit=max&auto=format&n=pPcK1s6Qs7BTlBW2&q=85&s=cb2b7173992862e45da3113c20014e5c" href="/gateway/model/how-tos/connect-a-client" width="24" height="24" data-path="assets/integrations/openai.svg">OpenAI-compatible client setup</Card>
  <Card title="Anthropic" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/anthropic.svg?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=591969959aa9c63717931b83693c6e9f" href="/gateway/model/how-tos/connect-a-client" width="46" height="46" data-path="assets/integrations/anthropic.svg">Messages client setup</Card>
  <Card title="OpenRouter" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/openrouter.svg?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=c91b1338c1f36c1d0bb6255e06384908" href="/gateway/model/references/compatibility" width="300" height="300" data-path="assets/integrations/openrouter.svg">Model capabilities and limits</Card>
  <Card title="Amazon Bedrock" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/bedrock.svg?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=5a2a141b173520283420c7b9d5ebc42b" href="/gateway/model/references/compatibility" width="16" height="16" data-path="assets/integrations/bedrock.svg">Native provider translation</Card>
  <Card title="Google" icon="https://mintcdn.com/atlan-602e2b74/pPcK1s6Qs7BTlBW2/assets/integrations/google.svg?fit=max&auto=format&n=pPcK1s6Qs7BTlBW2&q=85&s=16fac321f7e3207c93f65cb041d9de20" href="/gateway/model/references/compatibility" width="118" height="120" data-path="assets/integrations/google.svg">Vertex AI and Google AI</Card>
  <Card title="Azure" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/microsoft.png?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=4527aa5fac60292b4f6926f20d1731d1" href="/gateway/model/references/compatibility" width="500" height="500" data-path="assets/integrations/microsoft.png">Azure resource endpoints</Card>
  <Card title="LiteLLM" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/litellm.svg?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=71d4c352e0619a2f4a30158bd8ee0ac3" href="/gateway/model/references/compatibility" width="372" height="372" data-path="assets/integrations/litellm.svg">Proxy catalog and inference</Card>
  <Card title="TypeSafe" icon="https://mintcdn.com/atlan-602e2b74/IySoseVxu4BPYuCl/assets/integrations/typesafe.png?fit=max&auto=format&n=IySoseVxu4BPYuCl&q=85&s=876636d6e153e8c185dc427275723a1e" href="/gateway/model/references/compatibility" width="400" height="400" data-path="assets/integrations/typesafe.png">Typed decisions</Card>
</CardGroup>

## Catalog administration

Organization administrators can use **Providers → Add a provider** in the Model Gateway studio. The form changes to the selected provider type. It collects the endpoint and credential fields that provider requires. Custom endpoints must be allowed by your deployment.

After connecting a provider, inspect **Models** before updating clients. Disable a provider to stop its models from routing; disable an individual model to stop only that model. Refresh discovery when the upstream catalog changes. Discovery does not prove that every listed model supports every API operation.

## Verify a connection

With the application's agent key, list models through the gateway, copy an exact `id`, and make one request for its modality. A successful upstream sign-in or provider discovery alone does not verify inference.

See [Connect a client](/gateway/model/how-tos/connect-a-client) and [Model Gateway errors](/gateway/model/troubleshooting).


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