Replacement Models for Retired Models supporting Assistant API

Baswaraj Mamidgi 20 Reputation points
2026-01-16T03:55:43.0066667+00:00

Hi, we are currently using Azure Assistant API for chat, with GPT-4o model. Recently we got a depreciation Notice that gpt-4o model version(s) 2024-05-13 and 2024-08-06 will be retired. What are the latest replacement models to be used with Assistant API?

https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?view=foundry-classic&tabs=global-standard-aoai%2Cglobal-standard&pivots=azure-openai#assistants-preview

Azure OpenAI gpt-4o model version(s) 2024-05-13 and 2024-08-06 will be retired 31 March 2026 You're receiving this notification because you're associated with one or more Azure subscriptions that use Azure OpenAI. On 31 March 2026, Azure OpenAI gpt-4o model version(s) 2024-05-13 and 2024-08-06 will be retired. As part of this retirement, the replacement model will be set to gpt-5.1 version 2025-11-13 on 9 March 2026. If you've selected the option to auto upgrade your standard, data-zone standard, or global standard deployment(s), they will be automatically upgraded to the replacement model version during the weeks prior to retirement. Learn more about setting your deployment upgrade options. After the retirement date, model gpt-4o version(s) 2024-05-13 and 2024-08-06 will no longer be available or operable. This retirement will only timpact base-model deployments. Beginning on 31 March 2026, fine-tuning on model gpt-4o version(s) 2024-05-13 and 2024-08-06 will no longer be allowed. Existing fine-tuned deployments will continue to operate for an additional year. Required action To avoid service disruptions, please follow these instructions to migrate any apps, workloads, or dependent processes that these models use by redeploying them to other supported models by 31 March 2026. Get guidance on evaluating and upgrading models.

Azure AI services
Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
0 comments No comments
{count} votes

Answer accepted by question author
  1. Sridhar M 4,910 Reputation points Microsoft External Staff Moderator
    2026-01-16T04:20:10.69+00:00

    Hi Baswaraj Mamidgi

    Azure OpenAI has officially announced the retirement of GPT‑4o base model versions 2024‑05‑13 and 2024‑08‑06. These versions apply to base (non‑fine‑tuned) deployments, including those consumed through the Assistants API (Preview). After the retirement date, these GPT‑4o versions will no longer be deployable, callable, or operable in Azure OpenAI.

    Retirement timeline and effective dates:

    The final retirement date for GPT‑4o (2024‑05‑13 and 2024‑08‑06) is 31 March 2026. Azure subscriptions received formal notifications starting March 2026. From 31 March 2026 onward, any Assistant, chat, or workload pointing to these versions will fail unless migrated.

    Fine‑tuning impact As of 31 March 2026, new fine‑tuning on GPT‑4o (2024‑05‑13 / 2024‑08‑06) is no longer allowed. However, existing fine‑tuned GPT‑4o deployments are allowed to continue running for an additional one‑year grace period (to help customers avoid sudden production outages). Even during this grace period, Microsoft strongly recommends migrating as early as possible.

    Official replacement model for Assistants API:

    The official replacement model for GPT‑4o in the Assistants API (Preview) is gpt‑5.1 version 2025‑11‑13. This model is the supported successor for agent‑style, tool‑using, threaded Assistant workloads, and it fully replaces GPT‑4o for this use case.

    Assistants API compatibility gpt‑5.1 (2025‑11‑13) is fully supported by the Azure OpenAI Assistants API (Preview). It supports:

    • Threaded conversations
    • Tool and function calling
    • File retrieval / vector context
    • System + developer + user instruction layering

    No migration away from the Assistants API itself is required—only the model deployment changes.

    Upgrade behavior and requirements:

    Azure does not auto‑upgrade GPT‑4o Assistant deployments to GPT‑5.1 by default. Customers must manually create a new deployment using:

    • Model: gpt‑5.1
    • Version: 2025‑11‑13

    You must then update your Assistant configuration or deployment reference to point to the new deployment. No Assistants API code changes are normally required.

    Behavioral differences to expect When migrating from GPT‑4o to GPT‑5.1, expect stricter system‑message enforcement, more deterministic tool‑calling, and slightly different reasoning and verbosity patterns. Testing is strongly recommended for:

    • Tool/function schemas
    • JSON‑constrained outputs
    • RAG and retrieval flows

    These are behavior changes, not breaking API changes

    Note:After 31 March 2026, GPT‑4o (2024‑05‑13 / 2024‑08‑06) deployments will stop working completely. Assistants referencing these deployments will return errors, and new fine‑tuning will remain blocked.

    Recommended migration steps

    1. Create a new deployment using gpt‑5.1 (2025‑11‑13)
    2. Update Assistants to reference the new deployment
    3. Validate tool calls, retrieval, streaming, and system instructions
    4. Decommission GPT‑4o deployments before 31 March 2026

    Reference:

    I Hope this helps. Do let me know if you have any further queries.


    If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

    Thank you!

    1 person found this answer helpful.

0 additional answers

Sort by: Most helpful

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.