A reasoning model trained with reinforcement learning, on par with OpenAI o1 — Replicate | AllCraft | Allcraft AI