- Moonshot has launched a new AI model, Kimi K3
- It’s surprisingly powerful, and the Chinese AI company claims it outperforms most of its American rivals, with a couple of exceptions.
- Kimi K3 is open weight in nature, posing an additional threat to companies like OpenAI and Anthropic.
Moonshot, one of the emerging Chinese AI giants, has just revealed a new AI model that is apparently on par with ChatGPT and Claude.
Bloomberg reports that Moonshot’s new Kimi K3 model can match the best the United States has to offer, at least according to the company’s own benchmarking. It apparently outperforms all rival AIs except Claude Fable 5 (from Anthropic) and GPT-5.6 (from OpenAI).
Kimi K3 is a model with 2.8 trillion parameters, Bloomberg tells us, and Artificial Analysis ranked it ahead of Anthropic’s Opus 4.8 in some benchmarks.
Moonshot also claims to outperform its Chinese rival Z.AI in encoding tasks, and overall the performance of the new model has taken the market by surprise.
Moonshot notes in a blog post that the Kimi K3 is the “world’s first open 3T class model, designed for frontier intelligence through coding, knowledge work, and long-term reasoning.”
Kimi K3 is now available for use. Bloomberg quotes Leonid Mironov, portfolio manager at Gavekal Capital, as saying: “In my use, it is clearly the best Chinese model of all time,” noting that it is nothing short of “brilliant.”
Analysis: a serious threat
This is a threat to the big American players in the AI market for several reasons.
The key difference with the Kimi K3 is that it is what’s known as an “open weight” model, meaning anyone can take the model and run it themselves (as of July 27, when weights are released) without paying anything. However, it’s not the same as open source, since while you can get the model, what you can’t do is take a look behind the scenes at how the model was trained (and with what data).
The other caveat is that running Kimi K3 requires extremely powerful hardware; However, for companies with the substantial means to do so, the pre-trained model is available at no cost. So you can imagine how this open weight approach is a threat to the AI giants in the US (and this is presumably the goal of going down this path for the Chinese rival).
What will also be a concern for companies like OpenAI and Anthropic is that Moonshot is attacking one of the most lucrative aspects of AI, with Kimi K3 coming under pressure for his coding skills. However, Moonshot charges much more than its Chinese rivals for those who want to use it and, in fact, is priced around Claude Sonnet levels, putting it on par with current cutting-edge (frontier) AI models.
That in itself is a sign of the quality on offer here, and why the Kimi K3 has caught the eye. As competition around AI intensifies, there are also concerns about whether this means safeguards will increasingly be overlooked in favor of faster development and progress (which has already been a constant source of concern for many).
Adding to all the controversy are accusations of AI theft leveled by the US State Department against Chinese companies earlier this year, including Moonshot.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
The best laptops for all budgets
