A new Moonshot AI release may narrow the gap with OpenAI and Anthropic, hinting at a major shift in enterprise AI choices.
The latest version of Moonshot AI’s Kimi line is expected to significantly close the gap to closed solutions from OpenAI and Anthropic, notably Opus 4.8, and in some features may even surpass them.
Kimi K2 models have already received positive feedback in the open-source market: they demonstrate strong benchmark standings and capabilities close to the best models.
It is expected that Kimi K3 will become the most sophisticated open-weight model from China; its parameter count is estimated to be between 2 and 3 trillion, and the release is planned in the coming days.
Context and Prospects
According to the same publication, Moonshot is preparing a new funding round that could lift the company’s valuation to $31.5 billion. In May, the company raised $2 billion at a valuation of around $20 billion.
Discussions about the cost of using closed models are intensifying amid concerns about how large AI labs handle customer data for their products, such as ChatGPT and Claude.
Executives also propose alternatives or recommend using more accessible open models from DeepSeek, Z.ai or Moonshot, tailoring them to their companies’ needs.
Overall, the trend points to increasing importance of open models in the global AI ecosystem and a desire among businesses to reduce dependence on closed platforms.