Chinese AI startup DeepSeek has introduced a preview of its latest model, V4, signaling its continued push to challenge global leaders in artificial intelligence with high-performance, low-cost systems.
The new V4 model, particularly its pro version, has demonstrated strong performance in world-knowledge benchmarks, outperforming several open-source competitors and trailing only Gemini-Pro-3.1, a closed-source model developed by Google.
DeepSeek, backed by High-Flyer Capital Management, gained global attention last year for its cost-efficient AI models that disrupted expectations around the economics of large-scale AI development.
The company is now reportedly seeking to raise fresh funding at a valuation exceeding $20 billion, with Chinese tech giants Alibaba and Tencent said to be in discussions to take stakes.
However, the launch comes amid growing geopolitical tensions around AI development. U.S. officials have accused China of systematically acquiring intellectual property from American AI labs, raising concerns over fair competition and technology security.
In response, China has rejected the allegations, stating its commitment to protecting intellectual property rights and dismissing the claims as unfounded.
The timing of the V4 preview highlights the intensifying global race in artificial intelligence, where performance, cost efficiency, and geopolitical considerations are increasingly intertwined.
As competition heats up, DeepSeek’s latest move reinforces its ambition to emerge as a major player in the global AI landscape, even as scrutiny around the sector continues to deepen.
