Retired model

OpenAI

At shut down, OpenAI says the model or endpoint is no longer accessible. Sunset means the same thing. The deprecations page does not print an HTTP status for that moment.

Source: OpenAI deprecations , checked October 22, 2026.

Anthropic

Retired means the model is no longer available for use. Requests to retired models fail. Requests past the retirement date fail. The deprecations page does not name an HTTP status for that failure.

Source: Anthropic model deprecations , checked October 22, 2026.

Gemini API and Vertex AI

The Gemini deprecations page says that once a model is shut down it is completely turned off and the endpoint is no longer available. It does not print one error sentence for every model.

Firebase's model page says Gemini 1.0 and Gemini 1.5 are already shut down and that requests to those models return a 404. That sentence is about those generations.

Vertex puts past models in its retired listing. A listed retirement date will not be moved earlier.

Sources: Gemini API deprecations , Firebase AI Logic models , Vertex model versions , checked October 22, 2026.

Amazon Bedrock

Bedrock calls this End-of-Life, not Retired. After the EOL date the model is removed and requests fail, unless you have a private arrangement with the provider. Migration does not happen automatically. On the legacy page, state can differ by Region.

Sources: Bedrock model lifecycle , Bedrock legacy lifecycle , checked October 22, 2026.

Azure OpenAI

Azure's lifecycle table says Retired means removed from service and all inference requests return 410 Gone. That status code is on the Azure page. It is not what the OpenAI deprecations page prints for the OpenAI API.

If the deployment's upgrade option is NoAutoUpgrade, the deployment stops working at retirement. Other upgrade options move the deployment onto a newer version instead.

Sources: Foundry Models lifecycle , Working with Azure OpenAI models , checked October 22, 2026.