# OpenAI Launches Model Distillation Feature to Cut AI Costs

OpenAI announced a new Model Distillation feature in its API that allows developers to create cheaper, faster AI models by training smaller models on the outputs of larger ones.

The feature works by letting users fine-tune cost-efficient models using responses generated by OpenAI's more expensive frontier models like GPT-4. This process, called distillation, transfers knowledge from a powerful "teacher" model to a smaller "student" model—all within OpenAI's platform.

**Why it matters:** Running large AI models is expensive. A single API call to GPT-4 costs significantly more than using smaller models like GPT-3.5. Model distillation offers a middle ground: developers can capture much of GPT-4's capabilities in a smaller model that costs less to run at scale.

This is particularly valuable for companies deploying AI in