DeepSeek R1

Overview

DeepSeek R1 is an open-weight reasoning model trained with large-scale reinforcement learning. It matches or approaches frontier closed models on math and code benchmarks while being free to self-host, fine-tune, and distill under the MIT license.

Key features

• Open weights under MIT license
• RL-trained chain-of-thought reasoning
• Distilled variants (1.5B–70B)
• 671B MoE with 37B active parameters
• Very low API pricing

Tags