Astra repeatedly returns "Selected model is at capacity" in Codex desktop and web on Pro

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

26.908.70816

What subscription do you have?

ChatGPT Pro — signed in with a ChatGPT account, not an API key.

What platform is your computer?

macOS

What issue are you seeing?

I am unable to use Astra reliably with my ChatGPT Pro subscription. Requests repeatedly fail with:

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

The desktop app also repeatedly reconnects and sometimes ends with:

"stream disconnected before completion: stream closed before response.completed"

The capacity error occurs in both the Codex desktop app and the official web interface. I switched from my company network to a mobile hotspot, but the same issue persisted.

The problem has continued for at least a day and is severely disrupting my daily work. I do not know whether the underlying cause is model capacity, account entitlement, or another service issue.

What steps can reproduce the bug?

The following reproduces the issue on my account:

1. Sign in to the Codex desktop app with a ChatGPT account that has a Pro subscription.
2. Select Astra and submit a request.
3. Observe the error: "Selected model is at capacity. Please try a different model."
4. Retry and observe repeated capacity errors, reconnection attempts, or a stream-disconnection error.
5. Switch from the company network to a mobile hotspot.
6. Sign in to the official web interface with the same account, select Astra, and submit a request.
7. Observe the same capacity error on the web interface.

This reproduces across both interfaces and both tested networks.

What is the expected behavior?

Requests to Astra should complete when access and capacity are available.

If access is temporarily unavailable, the app should provide a clear, actionable explanation and recover when the issue is resolved, rather than repeatedly cycling through capacity errors and reconnection attempts.

If the issue is related to my account or subscription entitlement, I would appreciate guidance on how to resolve it.

Additional information

I am a long-time Codex user and rely on it for daily work. This ongoing issue has brought much of my work to a standstill.

I contacted OpenAI Support through the Help Center and received confirmation that my case was escalated to a support specialist. After a full day, the issue remains unresolved and I have not received a substantive update.

I can provide account-specific information privately through my existing support case. Please let me know what non-sensitive diagnostic information would help investigate this issue.

Thank you for reviewing this report.

The report names no repository files, tests, or entry points. Start by determining whether the failure is account entitlement, service capacity, or shared behavior across the desktop and web interfaces; done requires an identified cause and a confirmed resolution rather than another client-side retry.