DeepSeek, a company based in China, has made an improved version of its R1 artificial intelligence model available to developers using Hugging Face. The news came in a message posted by DeepSeek on WeChat early Wednesday.
Users can access the newest R1 model under an MIT license, which allows for commercial use. According to DeepSeek, this release is a small upgrade from the earlier edition.
Although DeepSeek uploaded the model to Hugging Face, only technical files and the model’s essential weights were included, with no extra description or usage guide for developers. The model’s scale is striking, with 685 billion parameters powering its reasoning abilities.
Experts note that ordinary computers at home are unlikely to handle the sheer computational needs of such a large model without special modifications.
DeepSeek’s Rapid Rise in Artificial Intelligence
This fast-growing firm first drew widespread attention when its initial DeepSeek R1 model release matched the performance of tools from leading American firms in the field. Some regulators in the United States have expressed concern, claiming DeepSeek’s technology could present a national security threat.
Despite these concerns, the popularity of DeepSeek’s models continues to rise among those seeking advanced artificial intelligence tools. China’s role as a growing force in AI development is under the spotlight as companies like DeepSeek push technical boundaries, showcasing key China AI advancements.
