codex - 💡(How to fix) Fix codex imagegen bug 无法生图

What version of the Codex App are you using (From “About Codex” dialog)?

版本 26.602.30954

What subscription do you have?

codex imagegen 无法生图

What platform is your computer?

macOS (Apple Silicon)Mac Studio

What issue are you seeing?

昨天还都正常，今天点了升级最新版本后，生成图片，一直说限流无法生成，偶尔能成功一张，这样根本就没法用了啊。还有人也出现这个情况吗？？What issue are you seeing?

I am using the official Codex Desktop app and signing in through ChatGPT (not using a local OPENAI_API_KEY).

When I ask Codex to generate images using the built-in image generation tool (imagegen skill), image generation sometimes fails immediately with:

TooManyRequests

Encountered exception: <class ‘oai_errors.errors.TooManyRequests’>

image generation request was rate limited

The issue is confusing because my Codex usage page still shows significant remaining quota. At the time of the error:

• 5-hour usage remaining: 67%
• Weekly usage remaining: 91%

In addition, image generation works normally in ChatGPT using the same account.

This does not appear to be:

• a content policy issue
• a prompt safety issue
• a local API key issue
• an account quota exhaustion issue

The Codex app provides no explanation about whether the rate limit is caused by:

• image-specific quotas
• backend service load
• temporary cooldown
• account restrictions
• or a known service issue

The error message provides no retry time or recovery guidance.

Steps to Reproduce

1. Open Codex Desktop.
2. Sign in with ChatGPT.
3. Start a new conversation.
4. Ask Codex to generate an image, for example: “Generate an image of a red apple on a white background.”
5. Codex may immediately return: TooManyRequests Encountered exception: <class ‘oai_errors.errors.TooManyRequests’> image generation request was rate limited

Expected Behavior

Codex Desktop should either:

1. Successfully generate images using the built-in image generation capability; or
2. Provide a clear user-facing explanation including:
   • whether the rate limit is related to image generation specifically
   • whether it is caused by backend service load
   • whether it is a temporary cooldown
   • estimated retry timing
   • whether the displayed Codex usage quota includes image generation usage

Additional Information

The issue occurred during normal image generation tasks and was not related to coding workflows.

The account still had substantial remaining Codex usage quota when the error occurred.

The issue later resolved itself without any configuration changes, suggesting that the error may be related to backend image generation service availability or image-specific rate limiting rather than actual account quota exhaustion.

Improved error messaging would help users understand whether they should wait, retry later, or take other actions.

What steps can reproduce the bug?

1. Open Codex Desktop.
2. Sign in with ChatGPT.
3. Start a new conversation.
4. Ask Codex to generate an image, for example: “Generate an image of a red apple on a white background.”
5. Codex may immediately return: TooManyRequests Encountered exception: <class ‘oai_errors.errors.TooManyRequests’> image generation request was rate limited

What is the expected behavior?

No response

Additional information

No response