Fine Tuning or similar

I have seen a few questions around here about this but no responses to the questions.

I have an interest in being able to fine tune the models used by Cursor. I have a project that uses its own extension of React Native and all of the extensions are unknown when using Cursor.

I do appreciate that fine tuning is generally done across the LLM being used (for example 4o or something like that). And, as such, fine tuning would lock a team to a specific model/implementation that has been tuned. So, this becomes a pretty specific use case. But for enterprises this can be a full requirement.

OpenAI (and perhaps all others by now) allows Fine Tuning. So, this may be more a case of “can we use a special model from OpenAI” question.

Unfortunately, we don’t have any facility for users to fine tune the models in the traditional sense. We do a lot of work and magic behind the scenes on the built in models so that they work smoothly inside Cursor, and giving the option to find-tune them yourself would likely break this.

As a workaround, you can provide detailed instructions to the AI on how to behave, but this will only get you so far compared to fine tuning!