OpenAI o1 (formerly Strawberry) — Reasoning Models Arrive

OpenAI releases o1, a new model class trained with reinforcement learning on reasoning. Chain-of-thought becomes a core, trained-in capability. Math benchmarks leap.