Openrouter api support

give me a hour

you can use the env var OPENAI_BASE_URL

you can use the env var OPENAI_BASE_URL

This doesn't seem to work right now even if you set up the baseurl and model names correctly since the agent loop requests the chat completions using the new openai /v1/responses api instead of the older /v1/chat/completions api which other providers already support.

This however seems to be a deliberate way to discourage using your own models to run codex and some work on this front would be very helpful for all.

So maybe adding something like this proxy to the mix:

https://github.com/open-responses/open-responses

Should we keep this titled as OpenRouter API support, or /chat/completions instead? If the latter, it basically opens up a floodgate of options.

Is anyone still working on this? Would love to use openrouter on codex

Looks like there's a fork here https://github.com/ymichael/open-codex, but most discussion about this feature is in https://github.com/openai/codex/issues/26

Thank you for the feedback! Please try again on the latest version of Codex CLI. If the issue persists, let us know.

@codex-maintainers
This feature does not work on the latest version of codex.

> sudo npm i -g @openai/codex
> codex --version
codex-cli 0.30.0

<img width="534" height="122" alt="Image" src="https://github.com/user-attachments/assets/af421ac7-fb8a-4d74-a44d-9fa405a28aae" />

Found the solution, looks like the way to use other models has been changed, one needs to refer to this doc to use other models(by using config.toml): https://github.com/openai/codex/blob/main/docs/config.md

This is a working config.toml to use openrouter if anyone need

model_provider="openrouter"
model="openai/gpt-5"

[model_providers.openrouter]
name="openrouter"
base_url="https://openrouter.ai/api/v1"
env_key="OPENROUTER_API_KEY"

Found the solution, looks like the way to use other models has been changed, one needs to refer to this doc to use other models(by using config.toml): https://github.com/openai/codex/blob/main/docs/config.md This is a working config.toml to use openrouter if anyone need `` model_provider="openrouter" model="openai/gpt-5" [model_providers.openrouter] name="openrouter" base_url="https://openrouter.ai/api/v1" env_key="OPENROUTER_API_KEY" ``

This still doesn't work for me. After setting above, you you run codex -model openrouter ?

> Found the solution, looks like the way to use other models has been changed, one needs to refer to this doc to use other models(by using config.toml): https://github.com/openai/codex/blob/main/docs/config.md > This is a working config.toml to use openrouter if anyone need > `` > model_provider="openrouter" > model="openai/gpt-5" > > [model_providers.openrouter] > name="openrouter" > base_url="https://openrouter.ai/api/v1" > env_key="OPENROUTER_API_KEY" > ` This still doesn't work for me. After setting above, you you run codex -model openrouter` ?

after updating the config, just run codex works for me

BTW, you will also need the OPENROUTER_API_KEY env in your .bashrc/ .zshrc

this config.toml ，It doesn't work for me，still need login

<img width="611" height="203" alt="Image" src="https://github.com/user-attachments/assets/6deadf43-3d1f-4083-aeba-e1c14842f083" />
codex-cli 0.136.0