What each integration covers
Use Microsoft Foundry for every new Azure model account. To move an existing Azure OpenAI or Azure AI Foundry account, follow Migrate to Microsoft Foundry. Hub and hub-based Azure resources cannot be migrated.
Microsoft Foundry
Microsoft Foundry uses the OpenAI-compatible/openai/v1 surface for compatible deployments, and provider-specific routes where a publisher needs them.
A Microsoft Foundry account can use:
- Chat completions, legacy text completions, and the Responses API, including compaction
- Embeddings, image generation, and image editing
- Text-to-speech, audio transcription, audio translation, and realtime sessions
- Files, batches, and fine-tuning
- Cohere rerank and Mistral Document AI OCR
- Anthropic Messages and token counting for Claude deployments
- Configure the resource’s OpenAI endpoint and, for discovery and provider-specific models, its project endpoint.
- Configure authentication once, on the provider account. API key, certificate, client secret, and workload identity federation are supported.
- Add integrations by Azure deployment name, or discover them from the project endpoint.
- Call a Gateway model ID such as
my-foundry/gpt-4.1.
Resource endpoints
Microsoft Foundry accounts can use two endpoint fields:- OpenAI endpoint: Used for the OpenAI-compatible
/openai/v1inference surface. - Project endpoint: Used for deployment discovery and resource-level provider routes such as Anthropic Messages and Cohere rerank.
Do not append
/openai, /openai/v1, or an API path to the configured endpoint. The Gateway constructs the upstream path.