Browse Models
The DeepSeek R1 family consists of large language models developed by Deepseek AI that utilize reinforcement learning and Mixture-of-Experts architecture to enhance complex reasoning capabilities. The models feature 671 billion total parameters with 37 billion active during inference, employ supervised fine-tuning followed by reinforcement learning training stages, and demonstrate competitive performance on mathematical, coding, and reasoning benchmarks with context windows up to 128,000 tokens.