DeepSeek-R1 (and distilled checkpoints) Open (MIT-licensed) reasoning model with distilled 1.5B–70B local checkpoints; known for chain-of-thought quality. 0390 Open-source Models# DeepSeek-R1# distilled# local