Which models we can use as a replacement of the gpt-4.1 and gpt-4.1-mini

Hello @Dhruvkumar Babariya (Persistent Systems Inc) ,

Welcome to Microsoft Q&A .Thank you for reaching out to us.

Based on the current Azure AI Foundry model retirement schedule, an official one-to-one replacement has not yet been published for either gpt-4.1 or gpt-4.1-mini, as the replacement column is currently blank for both models.

For migration planning, the recommended approach is to evaluate the latest supported GPT models available within the target Azure region and deployment type.

If migration validation is starting now, the primary models recommended for evaluation are:

• gpt-4.1 → Evaluate the latest available GPT-5 family model in the target region.
• gpt-4.1-mini → Evaluate the latest available GPT-5 mini-class model in the target region.
• gpt-4.1-nano (if applicable) → Evaluate the latest lightweight GPT model available for the deployment type.

To minimize risk and ensure a smooth transition, the following approach is suggested:

1. Please create a new deployment using the selected target model instead of modifying the existing deployment.
2. Retain the current gpt-4.1 or gpt-4.1-mini deployment during testing.
3. Execute representative workloads against both deployments and compare:
   • Response quality and reasoning behavior
   • Latency and throughput
   • Token consumption and cost
   • Function/tool calling behavior
   • Structured output consistency (for example JSON responses)
4. Where possible, configure applications to reference the deployment name rather than a hard-coded model version. This simplifies future model migrations and reduces application changes during upgrades.
5. Gradually transition production traffic after successful validation.
6. Complete migration before the retirement date to avoid service disruption.

The most suitable replacement model can be identified more precisely with the following information:

• Azure region hosting the Azure OpenAI resource
• Deployment type (Standard, Global Standard, Data Zone Standard, or Provisioned)
• Workload type (Chat Completions, Responses API, Assistants/Agents, etc.)
• Primary objective (quality, latency, throughput, or cost optimization)

Since model availability and quota allocation can vary by region and deployment type, these details help determine the most appropriate migration path.

These should be considered migration candidates rather than officially published successors and should be validated against workload requirements before adoption.

The following references might be helpful , please check them out

• Model retirement schedule - Microsoft Foundry | Microsoft Learn
• Foundry Models lifecycle and support policy - Microsoft Foundry | Microsoft Learn
• Foundry Models sold by Azure - Microsoft Foundry | Microsoft Learn
• Microsoft Foundry Models overview (classic) - Microsoft Foundry (classic) portal | Microsoft Learn

Please let us know if the response was helpful

Thank you