400 Bad Request: {"detail":"Unsupported model"} with gpt-5-mini/nano when being logged in with ChatGPT
Resolved 💬 9 comments Opened Aug 9, 2025 by kevinmarkwardt Closed Oct 27, 2025
💡 Likely answer: A maintainer (seratch, member)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.19.0
Which model were you using?
gpt-5-mini
What platform is your computer?
Ubuntu Linux
What steps can reproduce the bug?
Launch codex and specifying the gpt-5-mini model
~/workspace/$ codex --model gpt-5-mini-2025-08-07
>_ You are using OpenAI Codex in ~/workspace/
To get started, describe a task or try one of these commands:
/init - create an AGENTS.md file with instructions for Codex
/status - show current session configuration and token usage
/diff - show git diff (including untracked files)
/prompts - show example prompts
user
hello
🖐 unexpected status 400 Bad Request: {"detail":"Unsupported model"}
/status
📂 Workspace
• Path: ~/workspace/
• Approval Mode: on-request
• Sandbox: workspace-write
👤 Account
• Signed in with ChatGPT
• Login:
• Plan: Pro
🧠 Model
• Name: gpt-5-mini-2025-08-07
• Provider: OpenAI
• Reasoning Effort: Medium
• Reasoning Summaries: Auto
📊 Token Usage
• Input: 0
• Output: 0
• Total: 0
~/workspace/auto-coder$ codex --model gpt-5-mini
>_ You are using OpenAI Codex in ~/workspace/auto-coder
To get started, describe a task or try one of these commands:
/init - create an AGENTS.md file with instructions for Codex
/status - show current session configuration and token usage
/diff - show git diff (including untracked files)
/prompts - show example prompts
user
hello
🖐 unexpected status 400 Bad Request: {"detail":"Unsupported model"}
/status
📂 Workspace
• Path: ~/workspace/
• Approval Mode: on-request
• Sandbox: workspace-write
👤 Account
• Signed in with ChatGPT
• Login:
• Plan: Pro
🧠 Model
• Name: gpt-5-mini
• Provider: OpenAI
• Reasoning Effort: Medium
• Reasoning Summaries: Auto
📊 Token Usage
• Input: 0
• Output: 0
• Total: 0
~/workspace/$ codex --version
codex-cli 0.19.0What is the expected behavior?
It uses the model and the Codex works as normal
What do you see instead?
🖐 unexpected status 400 Bad Request: {"detail":"Unsupported model"}
Additional information
_No response_
9 Comments
Please block this spammer ^ previous comment. I reported
Can confirm this also happens with nano as well.
Hopefully they fix this soon as gpt-5-mini is much faster
More details on this issue: this can occur when you signed in with ChatGPT (ChatGPT backend currently supports only gpt-5). If you're fine with going with an api key,
/logoutthen set gpt-5-mini model (I know you all here need this way though).@seratch in ChatGPT we can switch between Fast and Thinking modes. How can we do the same in Codex?
Can we update this to a feature request instead of a bug, as I was unaware that it was not a supported option in the original gpt-5 release.
Guys this came as a surprise. I've been running my plus subscription in codex using gpt-5-mini that has been working flawlessly until this message started appearing today. I fail to understand the rationale, codex supporting chatgpt logins for plus users, but now limiting the models on codex when they are available on web. I appreciate the challenges with release but this is more about having consistent policy. Models available under the subscriptions should be supported on codex and web alike. =/ Just my two cents
P.S. As of now as a plus user, i can only user gpt-5 and codex-mini-latest on codex and not any other models which are available in my subscription.
Why is it considered a feature request in this case? The OpenAI codex page states all models in the api should be supported and that you can sign in with your plus account.
They should update their documentation first in my opinion.
Not to mention the models we want are cheaper to use, seems counterintuitive to do that.
FYI this solved my issue during the AWS outage today. If anyone is having this error code come up during a Codex CLI session, this fixes it if you're simply using default codex models.
There are several different points made in the thread above, so it's unclear to me whether this is still an issue. I'm going to mark this as closed. If you have related questions, feel free to post in the discussion forum. And if you think that you're still seeing a bug related to this issue, please open a new bug report.