Feature request for product/service
API Keys
Describe the request
Can we get a bit of love for custom endpoints support? As of now, if you want to use any v1 compatible endpoint (there are many providers with PAYG, as well as subscriptions like chutes/nanoGPT/cerebras), your only option is to use “Override OpenAI Base URL”, which breaks all openAI models. Can we get a separate field for that, so we can continue using openAI models trough cursor but any custom added models go through custom endpoint?
I just tested, I manually added GLM 4.5 as custom model, and it worked just fine in agent mode, so the functionality is there and it works already (but it disables openAI models, which are my main models for coding big things in cursor). I don’t really ask for complex redesign with per model endpoint choice (as implementing that sounds pretty annoying and time-consuming), but small feature like separate field for openAI compatible providers seems reasonably easy to do, considering that it already is basically supported?