Codex Desktop repeatedly shows “Selected model is at capacity” only on my Pro account

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

26.825.41651

What subscription do you have?

PRo

What platform is your computer?

Windows 11

What issue are you seeing?

I am experiencing a persistent account-specific issue with Codex Desktop on my ChatGPT Pro account.

On the same Windows PC, same Codex Desktop installation, same network, and same repository, other ChatGPT accounts work normally and can continue Codex tasks without problems. However, when I switch back to my Pro account, Codex repeatedly fails with:

“Selected model is at capacity. Please try a different model.”

I have tested GPT-5.6 Sol with Medium, High, and Very High reasoning levels, and I have also tried GPT-5.6 Terra. The problem persists across model/reasoning changes.

Sometimes Codex starts working normally for a short period, reads files, edits files, and executes commands, but then the task fails mid-run with the same capacity error. In some cases, repeated retries also fail.

I have already tested:

Restarting Codex Desktop
Starting new conversations
Changing models
Changing reasoning levels
Signing out and signing back in
Retrying after the cooldown
Testing other accounts on the exact same PC/network

The important part is that other accounts work normally under the same conditions, while my Pro account consistently experiences the capacity error. This suggests the problem may be related to account-level routing, capacity allocation, admission control, or backend state rather than my local environment.

My usage quota is not exhausted, and this does not appear to be a normal usage-limit message.

Environment:

Codex Desktop
Windows 11
ChatGPT Pro account
Models tested: GPT-5.6 Sol and GPT-5.6 Terra
Reasoning tested: Medium / High / Very High

Could you please check the backend routing/account state for my Pro account and investigate why it is receiving repeated model at capacity errors while other accounts on the same device work normally?

I can provide screenshots, timestamps, logs, and a Feedback ID if needed.

What steps can reproduce the bug?

Open Codex Desktop on Windows 11 and sign in with my ChatGPT Pro account.
Open any local repository/project and start a normal Codex task.
Use GPT-5.6 Sol with Medium, High, or Very High reasoning. I also tested GPT-5.6 Terra.
Codex may initially work normally: it can read files, edit files, and execute commands.
After a short period, or sometimes immediately, Codex repeatedly returns:

“Selected model is at capacity. Please try a different model.”

Retrying does not reliably resolve the problem. Changing the model or reasoning level also does not resolve it.
Restarting Codex Desktop, signing out/in, and starting a new conversation do not resolve it.
Most importantly, on the same Windows PC, same Codex Desktop installation, same network, and same repository, other ChatGPT accounts work normally. The problem consistently occurs on my Pro account.

The issue can also occur in the middle of an active agent task after Codex has already edited files or executed commands. My usage quota is not exhausted.

What is the expected behavior?

Codex Desktop should continue the active agent task normally without repeatedly failing with a model-capacity error.

If the selected model is temporarily unavailable, Codex should retry gracefully, queue the request, or route the task to available capacity instead of interrupting an already running task.

My Pro account should behave consistently with other accounts tested on the same device and network.

Additional information

No response

No repository file or test is identified. Start by reproducing the failure in Codex Desktop on Windows 11 with the Pro account, comparing the same PC, network, repository, models, and reasoning levels against other accounts; collect the offered logs, timestamps, and Feedback ID. Done means active tasks no longer receive repeated capacity errors and the Pro account behaves consistently under the same conditions.