No response from any model
Resolved 💬 4 comments Opened Oct 11, 2025 by GoodFarming Closed Oct 11, 2025
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.46.0 - tried rollback to 0.45.0 and 0.44.0
Which model were you using?
tried codex and gpt-5
What platform is your computer?
wsl ubuntu
What steps can reproduce the bug?
totally dead for me, no response - tried reinstall, rollback, system restart, system update, npm updae, node update, etc, etc.
nothing else to add, just dead.
<img width="679" height="444" alt="Image" src="https://github.com/user-attachments/assets/19d45fbb-80cc-420d-8464-415bd22892ea" />
What is the expected behavior?
_No response_
What do you see instead?
_No response_
Additional information
_No response_
4 Comments
switched to using API instead of login and still nothing
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
works in powershell on same machine
Summary (WSL Ubuntu): Codex “opens but no responses”, Ctrl+C ignored, then “Unsupported model”
Root cause: A mismatch between nvm and a custom npm prefix in ~/.npmrc created multiple Codex shims on $PATH (some installed under Node 22, others elsewhere). The CLI was running under Node 22, which caused streaming/hanging behavior in WSL. After correcting the environment and reinstalling under Node 20 LTS, Codex worked.
Symptoms
Codex launches but prints nothing / retries forever; Ctrl+C sometimes ignored.
Works in Windows PowerShell but not in WSL.
Diagnosis
type -a codex showed multiple paths (e.g., ~/.npm-global/bin/codex and ~/.nvm/versions/node/v22.../bin/codex).
~/.npmrc contained prefix= (and/or globalconfig=), which conflicts with nvm.
Running from /mnt/c/... exacerbated PTY/stream quirks; running from ~ was stable.
Fix
Remove prefix/globalconfig from ~/.npmrc:
sed -i.bak -E 's/^(prefix|globalconfig)=/# \0/' ~/.npmrc
npm config delete prefix || true
npm config delete globalconfig || true
Use Node 20 LTS with nvm:
source ~/.nvm/nvm.sh
nvm install 20
nvm use --delete-prefix 20
node -v
Reinstall Codex under Node 20 and ensure PATH resolves to that shim:
rm -rf ~/.config/codex ~/.codex
npm uninstall -g @openai/codex || true
npm install -g @openai/codex@0.46.0
type -a codex
codex --version
Run from ~ (not /mnt/c/...) and pick a supported model:
codex login
codex --model gpt-5 # or select via /model in the UI
Takeaway: The issue wasn’t a service outage; it was an environment conflict (nvm + npm prefix + Node 22) leading to the wrong Codex binary/Node combo in WSL. Standardizing on Node 20 via nvm, removing the npm prefix, and reinstalling Codex resolved it.