"Request blocked" while working normally in a codex session
Open 💬 6 comments Opened Jul 19, 2026 by tarruda
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex CLI is running?
codex-cli 0.144.6
What subscription do you have?
Plus
Which model were you using?
gpt-5.6-sol
What platform is your computer?
Darwin 25.5.0 arm64 arm
What terminal emulator and version are you using (if applicable)?
_No response_
Codex doctor report
What issue are you seeing?
TBH This seems some API safety filter and not a codex cli bug, but this is the only place I'm aware that I can use to report. I was working on llama.cpp with codex and suddenly saw this:
<img width="1687" height="650" alt="Image" src="https://github.com/user-attachments/assets/7e325c02-002e-4619-9d14-23e51d3885e1" />
What steps can reproduce the bug?
Seemed random.
What is the expected behavior?
_No response_
Additional information
I was working on llama.cpp parsing/encoding of deepseek v4.
6 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
same, thread ID: 019f79dd-c445-7bc0-b0fa-775e8f5fee8a
Same here. A potential workaround is to switch your model to 5.5, continue the conversation and then swicth back after it replies.
Same here, the model suggested working with DeepSeek v4, Gemma or Qwen, was doing a few rounds to test the responses, checked the ChatGPT app to see how it was doing and found out it's been blocked.
Compacting on 5.5 and then moving back to Sol did work, though.
I had this experience too. Whenever a request is blocked, continuing the conversation with 5.5 fixes it.
my suspicion is the risk classifier has its own rate limit, and when it is hit it will block any subsequent requests automatically