When using custom/local model_provider we still need to either login with ChatGPT or provide API key
Resolved 💬 3 comments Opened Oct 23, 2025 by danechitoaie Closed Oct 23, 2025
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.47.0
What subscription do you have?
API
Which model were you using?
gpt-5 / gpt-5-codex
What platform is your computer?
Darwin 24.6.0 arm64 arm
What issue are you seeing?
When using custom/local model_provider we still need to either login with ChatGPT or provide API key.
It works even if we add a bogus API key, like "ABC" but doesn't make sense to require one.
What steps can reproduce the bug?
Run codex with custom/local model_provider
What is the expected behavior?
We should have a new option "3. Use custom model provider" (which doesn't do anything, just uses the custom model provider defined in config) next to the current ones "1. Login with ChatGPT", "2. Use API key"
Additional information
_No response_
3 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Thanks for the bug report.
When you use a custom model provider, you shouldn't see the login flow unless the provider is configured to required OpenAI auth (
requires_openai_auth = true).I just attempted to repro the problem, and it appears to be working for me. I added the following to my
config.toml, relaunched codex. It bypassed the login screen and was able to use the model.Please review your config to make sure the custom model is being specified correctly. If you're still seeing a problem, let us know if there are any additional repro steps.
Yes, just tested again and seems to be working.
I'm not really sure what I did that prompted me to auth.
We can close this. Thank you.