Selected model is at capacity on every model despite available usage quota

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

Codex CLI 0.154.0

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Linux 7.0.0-31-generic x86_64 desconocido

What issue are you seeing?

Codex CLI is currently unusable because every model I select returns the same error:

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

This happens even though I still have significant usage quota available.

At the time of the issue, Codex showed approximately:

• 5h: 83% left
• Weekly: 34% left
• Context: 0% used
• Context window: 400K

The issue is not specific to GPT-5.6 Sol. I have switched between multiple available models and reasoning levels, but they all return the same capacity error.

A fresh conversation with 0% context usage also fails immediately.

codex doctor reports a healthy installation:

• 19 ok
• 0 warn
• 0 fail
• Authentication configured correctly
• WebSocket connected successfully (HTTP 101 Switching Protocols)
• OpenAI/ChatGPT inference endpoint reachable
• Local state databases healthy
• No proxy configured

This suggests the issue is not caused by local connectivity, installation, authentication, context exhaustion, or normal usage-limit exhaustion.

What steps can reproduce the bug?

1. Open Codex CLI on Ubuntu Linux.
2. Start a new session with a fresh context.
3. Select any available model, for example GPT-5.6 Sol High.
4. Send any simple prompt.
5. Codex immediately returns:

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

6. Use /model and switch to another available model.
7. Send another prompt.
8. The same error appears.
9. Repeat with other available models and reasoning levels. The error occurs with every model tested.

At the time of testing:

• Codex CLI: 0.154.0
• 5-hour usage: 83% left
• Weekly usage: 34% left
• Context usage: 0%
• Context window: 400K

The issue also occurs in completely new sessions with 0% context usage.

What is the expected behavior?

Codex should process requests normally while usage quota is still available.

If a specific model is temporarily at capacity, switching to another available model should allow the request to continue.

Instead, every available model I have tested returns:

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

This happens even in a fresh session with 0% context usage and significant usage quota remaining.

Additional information

Environment:

• Codex CLI: 0.154.0
• Subscription: ChatGPT Plus
• OS: Ubuntu 26.04 (resolute), 64-bit
• Kernel: Linux 7.0.0-31-generic
• Architecture: x86_64

I ran codex doctor while the issue was occurring.

Result:

19 ok · 1 idle · 0 warn · 0 fail

Important diagnostic results:

• Installation: consistent
• Authentication: configured correctly
• Local databases: healthy
• WebSocket: connected successfully (HTTP 101 Switching Protocols)
• OpenAI provider endpoint: reachable
• No proxy configured
• Codex version: current (0.154.0)

The error persists across different models, reasoning levels, and completely new sessions.

Because all tested models fail while usage quota remains available and codex doctor reports no warnings or failures, the issue may be related to service-side capacity, account routing, or model admission rather than my local installation or network connection.

No source file or test is named in the report. Start by reproducing the failure across fresh sessions and models on the stated Linux environment, then inspect the CLI path associated with model selection and the reported capacity error. Done means identifying and fixing a client-side cause, or documenting evidence that the failure is service-side.