The latest version of China’s AI research institute Moonshot AI’s Kimi model series is expected to perform as well as or better than Anthropic’s Opus 4.8, the Financial Times reported, citing anonymous sources.
Moonshot’s Kim K2 model has been well received in the open source AI market, ranking highly in benchmarks and demonstrating capabilities comparable to the latest Frontier models.
The company’s next release, Kimi K3, is said to take this a step further and close the gap with closed source models such as OpenAI and Anthropic. FT reports that Kimi K3 will be China’s largest open-weight AI model with 2 trillion to 3 trillion parameters and will be released “within days.”
Moonshot is also raising new funding in a round that is said to bring the company’s valuation to $31.5 billion. The company raised $2 billion in May at a $20 billion valuation.
The news comes amid renewed debate about the value of paying for expensive closed-source models from AI labs like OpenAI and Anthropic. Industry leaders are concerned that AI labs could somehow extract data submitted by customers for use in AI products like ChatGPT and Claude.
Executives are touting their products as alternatives or encouraging companies to adopt cheaper open source models like those developed by DeepSeek, Z.ai, and Moonshot and train them for their own purposes.
