Chinese startup Moonshot AI has unveiled its Kimi K3 model, claiming it rivals and in some areas surpasses leading U.S. AI systems like those from OpenAI and Anthropic. While Kimi K3 doesn't match the overall performance of Anthropic's Claude Fable and OpenAI's GPT 5.6 Sol, Moonshot reports it outperformed other tested models on benchmarks for coding and general agents. The Kimi K3 model boasts 2.8 trillion parameters, making it China's largest AI model to date.

This development intensifies the U.S.-China AI race. Chinese AI models are increasingly being adopted by Western companies due to their competitive pricing and closing performance gap. Bank of America analysts, led by Alex Liu, noted that K3 demonstrates significant gains for Chinese models despite hardware constraints, highlighting innovation in pre-training and architecture.

Founded in 2023, Moonshot AI has quickly become a major player, raising $2 billion at a $20 billion valuation, with backing from tech giants like Alibaba and Tencent. The announcement led to stock drops for Chinese AI rivals, including Z.ai, whose shares fell 28%. Liu commented that K3 "raises the capability ceiling for China AI models, shifting the burden of proof to other independent AI labs."