DeepSeek, a Chinese AI developer, launched its open-source reasoning model DeepSeek-R1, which contains 671 billion parameters and showcases superior performance to OpenAI's o1 model on significant benchmarks. The model scores 79.8% on AIME 2024 and 97.3% on MATH-500, and it excels in coding tasks, achieving a 2029 Elo rating. Available on Hugging Face under an MIT license, DeepSeek-R1 and its distilled versions present an economically advantageous option for enterprises, with operational costs up to 95% lower than OpenAI's offerings.
DeepSeek-R1 achieves a score of 79.8% Pass@1 on AIME 2024, slightly surpassing OpenAI-o1-1217. It attains an impressive score of 97.3% on MATH-500.
For engineering-related tasks, DeepSeek-R1 performs slightly better than DeepSeek-V3, which could help developers in real-world tasks.
DeepSeek-R1 is available on the AI development platform Hugging Face under an MIT license, allowing unrestricted commercial use.
While OpenAI and other US-based firms have a first mover advantage, China is investing heavily in AI to build its capabilities.
Collection
[
|
...
]