Chinese AI lab DeepSeek released R1 on January 20, 2025, an open-source reasoning model with 671 billion parameters (37 billion active). Trained primarily via reinforcement learning for under $6 million, R1 matched or exceeded OpenAI's o1 on math and coding benchmarks — 79.8% on AIME, 97.3% on MATH-500.
Released under the MIT License, it demonstrated that frontier-level reasoning could be achieved at a fraction of Western labs' costs. The release caused significant drops in US tech stocks.