for model selection i just select the respective gpt models like if i didnt have a basurl, no custom model again.just with 5.5 builtin model, the other seems to work(tested with 5.4 & 5.4mini)
btw @deanrie , i was wondering if the new codex auth integration is somehow against cursor guidelines, i read em and there isnt anything explicit against it, but i can understand if it was in fact against that, if it is a proble, then i will delete the new codex integration from the repo and rewrite git hystory to get rid of it for good.
But i would hope it is allowed, it probably would bring much more users to cursor that already have codex subscription, and they would still need to pay for a pro account at least.
Hey, this is the same bug that’s discussed in GPT-5.5 BYOK not working. For the GPT-5 family, Cursor converts the request body to the Responses API format input instead of messages, but it only rewrites the URL to /v1/responses for the native OpenAI provider. With Override OpenAI Base URL, things get out of sync, and the request either breaks or doesn’t reach your backend at all. gpt-5.4 and gpt-5.4-mini go through a different path and work.
This is a known issue, it’s being tracked, and there’s no ETA yet. Subscribe to that thread for updates.
Workarounds:
Stay on gpt-5.4 or gpt-5.4-mini with BYOK.
Use gpt-5.5 without Override OpenAI Base URL.
On your second question about the Codex auth integration, that’s a policy or ToS topic, so I can’t comment on it on the forum. Email [email protected] with details of what the integration does and the team can give you a clear answer based on the guidelines.
