DeepSeek Launches V3 AI Model With Major Performance Boost and MIT License

DeepSeek has rolled out an upgraded version of its artificial intelligence model, V3-0324, boasting significant improvements in reasoning and programming capabilities. The update, quietly released on GitHub, also marks a shift to the MIT open source license, enhancing its accessibility for developers worldwide.

The new version is already attracting attention for outperforming its predecessor in early tests and even rivaling other compact models like ChatGPT’s o3-mini. AI entrepreneur Paul Gauthier noted on X that DeepSeek’s latest model showcases superior thinking capabilities compared to similar lightweight models.

This upgrade follows the release of DeepSeek R1 in late January, which made headlines for delivering top-tier performance using older hardware and far smaller budgets than competitors. The breakthrough fueled optimism around China’s growing AI capabilities and triggered a wave of activity among the country’s tech giants.

Following DeepSeek’s lead, companies such as Baidu (NASDAQ:BIDU), Bytedance, Alibaba (NYSE:BABA), and Tencent (HK:0700) have accelerated the release of their own AI innovations. Just last week, Tencent introduced its Hunyuan T1 model, claiming comparable performance and cost-effectiveness to DeepSeek R1.

DeepSeek’s strategic updates are not only pushing the boundaries of open-source AI development but also reinforcing China’s position in the global AI race. With performance, accessibility, and affordability at the forefront, DeepSeek V3-0324 is shaping up to be a key player in the evolving AI landscape.