o1

The o1 series of models are trained with reinforcement learning to perform complex reasoning. o1 models think before they answer, producing a long internal chain of thought before responding to the user.

Published on 2025-09-11 by @swift-ai.