›_ Codex Issues
Codex Issues / #25781

ChatGPT Plus account cannot use Codex extension: gpt-5.3-codex is not supported

Resolved 💬 6 comments Opened Jun 2, 2026 by stplat Closed Jun 2, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of the IDE extension are you using?

26.527.60818

What subscription do you have?

ChatGPT Plus

Which IDE are you using?

VS Code

What platform is your computer?

Linux 6.17.0-29-generic x86_64 x86_64

What issue are you seeing?

Every request in the Codex VS Code extension fails with:

{"type":"error","status":400,"error":{"type":"invalid_request_error","message":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}}

Additional details:

  • Active ChatGPT Plus subscription.
  • Codex usage quota is displayed correctly (99% remaining).
  • Authentication succeeds.
  • Codex is listed as included in ChatGPT Plus on the pricing page.
  • The issue started suddenly without local configuration changes.
  • Reinstalling / downgrading the extension does not appear to resolve the issue.

Expected behavior:
Codex should work for ChatGPT Plus users or automatically fall back to a supported model.

Actual behavior:
The extension appears to request gpt-5.3-codex and every request fails with a 400 error, making Codex unusable.

What steps can reproduce the bug?

<img width="456" height="2023" alt="Image" src="https://github.com/user-attachments/assets/096493a0-f946-40a2-9a4d-54fcfa0e36c1" />

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

6 Comments

github-actions[bot] contributor · 1 month ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #24856
  • #25440

Powered by Codex Action

CptHolzschnauz · 1 month ago

Same here on MacOS / VS Code

stplat · 1 month ago

Found the cause.

My ~/.codex/config.toml contained:

model = "gpt-5.3-codex"

The extension was using this pinned model instead of selecting a supported model automatically.

After removing the pinned model from the config (or renaming the config file), Codex started working again.

This appears to be related to config migration / model pinning after an update.

CptHolzschnauz · 1 month ago

Can't confirm, issue remains in my case

jvanakker · 1 month ago
Found the cause. My ~/.codex/config.toml contained: model = "gpt-5.3-codex" The extension was using this pinned model instead of selecting a supported model automatically. After removing the pinned model from the config (or renaming the config file), Codex started working again. This appears to be related to config migration / model pinning after an update.

I also have model = "gpt-5.3-codex" in my ~/.codex/config.toml, because I want gpt-5.3-codex to be the default. Removing it just causes it to switch to 5.5 by default, which does not solve our problem.

Conclusion: it just seems the gpt-5.3-codex model isn't available suddenly (unannounced). I think it's a bug, they normally announce phasing out models.

[EDIT] It seems they actually retired gpt-5.3-codex. Apparently it was announced a short while ago by Tibo on Twitter, no official news article at all on the OpenAI site. This is pretty wild, don't know of a good balanced replacement model that is as token-efficient as 5.3-codex.

CptHolzschnauz · 1 month ago

I removed the model = line as well, restarted, VS Code, old chats are wasted, the new chat says it's 5 codex with no subversion, I see no switch to choose a model anymore. Direction from OpenAI to GuidedAI ...