GPT-5.5 in Codex ignores configured context window and appears hard-capped around 258K tokens #19339

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

Latest

What subscription do you have?

Plus, Business

What platform is your computer?

No response

What issue are you seeing?

I set GPT-5.5 to 1,000,000 tokens, but the effective context is hard-capped at around 258,000 tokens.

This model is being presented for heavy-duty work, real work, and real workflows. If that is the case, why is it limited like this in practice?

What steps can reproduce the bug?

1. Open Codex and select GPT-5.5.
2. Set the context window to 1,000,000 tokens.
3. Start working until the session approaches the effective limit.
4. Observe that the usable context stops around 258,000 tokens instead of following the configured 1,000,000-token setting.

What is the expected behavior?

No response

Additional information

No response