o1

Definition

OpenAI's first reasoning model trained to think out loud before answering.

OpenAI's reasoning model post-trained to produce extended chains of thought before final answers.

Also called: o3, o3-mini

Mentioned in 4 episodes

1. 053
   An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
2. 041
   When the Iteration Teaches the Model to Skip the Iteration
3. 028
   Teaching a Model to Hire Copies of Itself: Recursive Agent Optimization
4. 026
   What RL Actually Does to Language Models, at the Token Level