Reasoning, released.
DeepSeek’s R1 release made model weights and a technical report available, with the main model released under the MIT license. Its approach uses reinforcement learning to develop reasoning capabilities.
DeepSeek reported results comparable to OpenAI’s o1 on selected mathematics, coding, and reasoning tasks. Those are the developer’s release-time evaluations, not a claim that one model leads every task today.
DeepSeek’s release and technical report ↗