GPT-3.5 Turbo fine-tuning and API updates

OpenAI launched supervised fine-tuning for GPT-3.5 Turbo, letting developers adapt the model to specific tasks and styles.

The feature aims to improve steerability, formatting, tone, and JSON output while reducing prompt length and API costs.

OpenAI said customer API data will not be used to train other models.

Support for function calling and GPT-3.5-Turbo-16k fine-tuning will arrive later, with GPT-4 fine-tuning planned for this year.