DeepSeek Releases Updated R1 Reasoning AI Model on Hugging Face

Chinese startup DeepSeek has unveiled an updated version of its R1 reasoning AI model on the developer platform Hugging Face, following an announcement via WeChat. The new version is licensed under the permissive MIT license, allowing for commercial use, and is described by DeepSeek as a “minor” upgrade.

The updated R1 model boasts a substantial size of 685 billion parameters, which indicates that it is quite powerful yet likely cannot be operated on typical consumer-grade hardware without alterations. The release of R1 earlier this year positioned DeepSeek as a strong competitor to models developed by OpenAI, drawing attention for its capabilities and also facing scrutiny regarding potential national security concerns from some regulators in the United States.