Skip to content

fix(ai): update OpenAI Codex model list#4603

Merged
badlogic merged 1 commit into
earendil-works:mainfrom
mattiacerutti:fix/openai-codex-model-list
May 17, 2026
Merged

fix(ai): update OpenAI Codex model list#4603
badlogic merged 1 commit into
earendil-works:mainfrom
mattiacerutti:fix/openai-codex-model-list

Conversation

@mattiacerutti

Copy link
Copy Markdown
Contributor

For the costs of the new models i've used the models.dev data.

I also had to update the tests since they were referencing a model which was removed.

Resolves #4601

@badlogic badlogic merged commit a01cf7a into earendil-works:main May 17, 2026
2 checks passed
@badlogic

Copy link
Copy Markdown
Collaborator

cheers!

@xHeaven

xHeaven commented May 17, 2026

Copy link
Copy Markdown

@mattiacerutti @badlogic When I'm trying to use gpt-5.5-fast through my ChatGPT subscription (Pro 5x), I'm getting this error:

Error: Codex error: {"type":"error","status":400,"error":{"type":"invalid_request_error","message":"The 'gpt-5.5-fast' model is not supported when using Codex with a ChatGPT account."}}

However, I can use GPT 5.5 Fast inside the Codex app/CLI. Am I doing something wrong, or is this expected for some reason?

@mattiacerutti

mattiacerutti commented May 17, 2026

Copy link
Copy Markdown
Contributor Author

Known issue, already addressed on main (see #4649)

@xHeaven

xHeaven commented May 18, 2026

Copy link
Copy Markdown

Known issue, already addressed on main (see #4649)

In this case, "addressed" being removed. Unfortunate. Thanks for the heads-up!

@mikker

mikker commented May 18, 2026

Copy link
Copy Markdown

Not exactly a model on its own but I've made Codex' /fast mode work by sending "priority" flag which Pi got from codex' own source

https://github.com/mikker/dotfiles/blob/e9c4155f3038da717f8d9ea5a5e771a1363f9f4e/pi/extensions/codex-fast.ts#L49-L57

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

certain gpt models throw unsupported errors when used through codex subscription

4 participants