Codex Desktop: All models repeatedly fail with "Selected model is at capacity" after ~10 seconds #53656

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

26.1007.21434

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Windows (64-bit)

What issue are you seeing?

I am experiencing a critical issue with the Codex Windows desktop application.

Error message:
"Selected model is at capacity. Please try a different model."

Problem description:

Almost every time I use Codex, the application fails approximately 10 seconds after starting a task.

This issue affects ALL available models. Switching to a different model does not resolve the problem.

As a result, Codex is almost completely unusable, and I cannot reliably complete my programming tasks.

Expected behavior:
Codex should execute coding tasks normally without repeatedly failing due to model capacity errors.

Actual behavior:
Tasks repeatedly fail with the same capacity error, regardless of which model is selected.

Impact:
This is a severe service availability issue that significantly disrupts my work.

Please investigate whether this is caused by server capacity, request routing, or an account-specific issue.

I would appreciate an urgent investigation and a permanent fix.

What steps can reproduce the bug?

1. Open the Codex desktop application on Windows.
2. Open an existing project or start a new conversation.
3. Select any available model.
4. Submit a coding task.
5. Wait approximately 10 seconds.
6. The task fails with: "Selected model is at capacity. Please try a different model."
7. Switch to another available model and repeat.
8. The same error occurs again.

What is the expected behavior?

The Codex desktop application should work reliably across all supported models.

When I submit a coding task, Codex should process the request and complete the task normally without repeatedly displaying model capacity errors.

Switching between available models should not result in the same failure.

I expect stable and reliable access to Codex as part of my paid ChatGPT Plus subscription.

Additional information

This is a severe and persistent issue, not an occasional error.

All available models are affected. The error repeatedly occurs approximately 10 seconds after starting a task, making the Codex Windows desktop application almost completely unusable.

I have a paid ChatGPT Plus subscription, but this issue prevents me from reliably using Codex for my programming work.

Please prioritize investigating this problem and determine whether it is related to server capacity, request routing, or my account.

I would appreciate a timely fix and an official explanation of the cause.

If this disruption continues, I would also appreciate guidance on whether a subscription billing adjustment is available.
I have multiple ChatGPT accounts, but only ONE specific account is affected by this issue.

The Codex desktop application works normally when I log in with my other ChatGPT accounts on the same computer.

However, when I switch to the affected account, all available models repeatedly fail with the error:

"Selected model is at capacity. Please try a different model."

This strongly suggests an account-specific issue rather than a general problem with my computer or network.

Please investigate the service availability, model access, and request routing for the affected account.

Since this GitHub issue is public, please provide a secure way for me to share the affected ChatGPT account information privately with the support team.