DeepSeek-R1-Distill-Qwen-7B

A 7B open-weight reasoning model distilled from DeepSeek-R1, built on the Qwen2.5-Math-7B base. It preserves much of DeepSeek-R1's step-by-step reasoning capability at a fraction of the size, making it practical for local deployment and low-cost inference on math, code, and general reasoning tasks. Source: supabase.