Skip to content
EOL Watch
Search itemsOpen menu

AzureRetirement

Microsoft Foundry Models: fine-tuned gpt-4.1 version 2025-04-14 deployment retirement

Deployment retirement date for fine-tuned gpt-4.1 version 2025-04-14 in Microsoft's retirement schedule for Foundry Models: 14 Oct 2027. Microsoft's lifecycle policy for Foundry Models says that at deployment retirement, inference and deployment return error responses.

Retires , 14 Oct 2027

What breaks

Microsoft lists fine-tuned gpt-4.1 version 2025-04-14 under "Fine-tuned models" of "Azure OpenAI" in the section "Foundry Models sold by Azure" in its retirement schedule for Foundry Models. The date in the "Deployment retirement date" cell of this row is 14 Oct 2027.

  • Microsoft says fine-tuned models retire in two phases: training and deployment.
  • The "Training retirement date" cell of this row reads: "No earlier than 2027-04-14", with the footnote "For existing customers only. Otherwise, training retirement occurs at base model retirement."
  • The policy says that after a model is retired for training, it is no longer available for fine-tuning but any previously trained models remain available for deployment.
  • The policy says that at deployment retirement, inference and deployment return error responses.

Who is affected

Affected versions

  • 2025-04-14

What to do

  1. In its section on migration, Microsoft's lifecycle policy says not to wait for Microsoft to name an official replacement before you begin evaluating newer models.
  2. The policy says to evaluate candidate models against your own application, prompts, and representative data when you choose a target.
  3. Microsoft's lifecycle policy says to confirm API compatibility, feature support, deployment-type and regional availability, capacity, and quota before you migrate.

Replacement

Not stated by vendor.

Sources

Last checked against the first source in this list on .

All 124 items for Microsoft Foundry Models

Found an error? Report it

The text on this page is drafted with AI assistance from the vendor's text, checked against the official source and approved by a person before it is published. How items are written

Last checked · Added · Changelog for this item