OpenAI’s self-serve fine-tuning availability

Just got an email about OpenAI winding down fine-tuning?

OpenAI is winding down the fine-tuning API and platform. Existing active customers can continue running fine-tuning training jobs through [January 6, 2027](calendar:T2:January 6, 2027) , after which creating new training jobs will no longer be possible.

And

Inference on fine-tuned models will only be turned off once the underlying base model is deprecated.

Maybe this is a way to save money but it kind of hurts the main appeal of using OpenAI services in my opinion.