deepseek-r1 (reasoning) is a post-trained (not simply finetuned) deepseek-r3 (base model) by deepseek. qwen is a completely separate 3rd party model by alibaba.

deepseek-r1 (reasoning) is a post-trained (not simply finetuned) deepseek-r3 (base model) by deepseek. qwen is a completely separate 3rd party model by alibaba. deepseek distilled (finetuned on the outputs of) r1 into qwen. They also did the same with llama.