DeepSeek’s First-generation Reasoning Models

DeepSeek’s first-generation thinking designs, accomplishing performance comparable to OpenAI-o1 throughout math, code, and thinking tasks.

Models

DeepSeek-R1

Distilled models

DeepSeek group has shown that the thinking patterns of bigger designs can be distilled into smaller sized models, leading to better efficiency compared to the thinking patterns found through RL on small models.

Below are the models created by means of fine-tuning against numerous dense extensively utilized in the research study community using reasoning information created by DeepSeek-R1. The assessment results show that the distilled smaller sized dense designs perform exceptionally well on standards.

DeepSeek-R1-Distill-Qwen-1.5 B

DeepSeek-R1-Distill-Qwen-7B

DeepSeek-R1-Distill-Llama-8B

DeepSeek-R1-Distill-Qwen-14B

DeepSeek-R1-Distill-Qwen-32B

DeepSeek-R1-Distill-Llama-70B

License

The model weights are accredited under the MIT License. DeepSeek-R1 series assistance commercial use, permit any adjustments and acquired works, consisting of, but not limited to, distillation for training other LLMs.