Moonshot AI to Launch Kimi K3 with 2–3 Trillion Parameters

Huo Xing Cai Jing reports, according to the Financial Times, that sources familiar with the matter have revealed that Chinese AI unicorn Moonshot AI plans to launch its new large language model, Kimi K3, in the near future. The model features 2 to 3 trillion parameters, making it the largest AI model in China by parameter scale, with performance expected to surpass Anthropic’s flagship model Claude Opus 4.8 (estimated by industry analysts to have 1.5 to 2 trillion parameters) in mainstream benchmark tests. Unlike current mainstream closed-source and expensive frontier models in the U.S., Kimi K3 will be released as an open-weight model, freely available for users to download and modify—potentially creating competitive pressure on U.S. leaders such as OpenAI and Anthropic. Amid rising service fees for U.S. large models—such as Anthropic’s recent announcement of a 50% price increase for Opus 4.8 in September—some overseas enterprises have begun shifting toward more cost-effective Chinese open-source models. In terms of capital markets, sources indicate that Moonshot AI is preparing a new funding round, with its latest valuation expected to reach approximately $31.5 billion. Meanwhile, valuations of other major AI players in China and the U.S. are also rising: DeepSeek is raising funds at a valuation of around $71 billion, while Anthropic and OpenAI have reached valuations of $965 billion and $852 billion, respectively, in their latest funding rounds. Moonshot AI has declined to comment on these rumors regarding the model launch and funding.