Azure provider: config.toml model changed after gpt-5.1-codex release, causing DeploymentNotFound errors
Resolved 💬 5 comments Opened Nov 14, 2025 by jgador Closed Dec 23, 2025
What version of the VS Code extension are you using?
0.4.40
What subscription do you have?
Azure OpenAI in Azure AI Foundry Models
Which IDE are you using?
VS Code
What platform is your computer?
Microsoft Windows NT 10.0.26200.0 x64
What issue are you seeing?
When using Codex with model_provider = "azure", Codex appears to rewrite my configured model = "gpt-5-codex" after the gpt-5.1-codex release. As a result, calls to Azure OpenAI start failing with DeploymentNotFound, even though the deployment exists and was previously working.
What steps can reproduce the bug?
- In Azure AI Foundry / Azure OpenAI, deploy gpt-5-codex.
- Configure Codex as shown above with model = "gpt-5-codex" and model_provider = "azure" and the rest of configuration following https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/codex?tabs=npm
- Use Codex in VS Code
- Observe that Codex fails with a 404 coming from Azure:
unexpected status 404 Not Found: {"error":{"code":"DeploymentNotFound","message":"The API deployment for this resource does not exist. If you created the deployment within the last 5 minutes, please wait a moment and try again."}}
What is the expected behavior?
_No response_
Additional information
_No response_
5 Comments
Please don’t deprecate gpt-5-codex. The gpt-5.1-codex model currently has serious regressions and fails on tasks that the previous version handled correctly.
@Beowulfwar thanks for letting me know. I didn't know about the regression on gpt-5.1-codex.
I still have a gpt-5-codex deployment working through AI Foundry. When 5.1 came out, the VS Code extension auto updated the config.toml file to gpt-5.1-codex so I got the same failure.
All I did was change it back to gpt-5-codex, reload the window, and back to work I went...for now!
@BSkando That's what I figured and did. I thought of filing issue here because I don't think they should be modifying the config that we explicitly set.
Azure continues to be an afterthought for the codex team and broader OpenAI team. I could see this being intentional to encourage us to use OpenAI services directly. If I was MSFT I'd be concerned/upset.