§01·model · /models
DeepSeek R1 Distill 14B
llmactiveMIT
14B reasoning model by DeepSeek — R1-style chain-of-thought distilled into a Qwen2.5-14B base, aimed at maths and code. Four cards carry a recipe here, all of them 24 GB or larger: the reasoning traces are long, and context is what costs the memory. MIT-licensed, which is unusually permissive for a reasoning model.
§02·GPUs that run this model
4 total✓ benchmarked·~ runs via recipe (not benchmarked)·— untested·✕doesn't fit