Chinese startup Moonshot AI on Friday released Kimi K3, a 2.8-trillion-parameter open-source model said to rival top US systems at a lower price, sending shares of Asian AI competitors into t
Chinese startup Moonshot AI on Friday released Kimi K3, a 2.8-trillion-parameter open-source model said to rival top US systems at a lower price, sending shares of Asian AI competitors into the red and reigniting the debate over who leads the global AI race.
The Beijing-based company unveiled the model this week and called it the largest open-source model released to date, according to Nikkei Asia and CNBC. Kimi K3 carries a 1 million token context window, enough to take in hundreds of pages at once, and handles both text and images natively.
Moonshot said the model is available now through Kimi.com, its mobile apps, and via an API, though the full model code will not be published until July 27, after which developers can then download, self-host, or make personal alterations.
In overall rankings, the Kimi K3 model sits below Anthropic’s Claude Fable 5 and OpenAI’s GPT 5.6 Sol, according to Yahoo Finance, trailing Fable 5 in several areas of evaluation. However, it beat Claude Opus 4.8 and GPT 5.5 on coding and general agent tests, with coding drawing the most attention to the Chinese model.
Arena.ai, a platform that ranks models on blind human comparisons, placed Kimi K3 ahead of Claude Fable 5 on its Frontend Code Arena leaderboard, according to Business Insider. In these blind tests, developers picked K3 over every leading US model on front-end coding tasks, and it matched GPT 5.6 Sol in Arena’s general text category.
Vercel CEO Guillermo Rauch wrote on X that it was “the first time that an open model is ahead of all proprietary ones for this comprehensive web engineering benchmark,” while warning that “benchmarks don’t always tell the full story.” Wharton professor Ethan Mollick called the model “closest to the frontier yet” while also urging caution regarding benchmark scores.
Most Chinese model launches have leaned on steep discounts. Moonshot has, however, taken a different path, setting API prices near Anthropic’s mid-range tier.
Kimi K3 costs $3 per a million input tokens and $15 per a million output tokens, with cheaper rates for cached inputs. OpenAI charges $5 and $30 for GPT-5.6 Sol, while Anthropic charges about $10 and $50 for Claude Fable 5.
This still leaves K3 as one of the cheaper frontier models, and pricing starts to really matter more as capabilities become harder to differentiate. A company running AI at scale can cut computing bills sharply by choosing a model that performs similarly for less.
The launch hit Asian AI stocks hard. Z.ai fell as much as 30% in Hong Kong trading, its worst single-day drop since listing in January. MiniMax Group dropped as much as 16%, while Alibaba, one of Moonshot’s backers, fell 4%. Bloomberg’s Asian semiconductor index saw a dip of more than 6%, and Nasdaq 100 futures fell 2%.
“K3 raises the capability ceiling for China AI models, shifting the burden of proof to other independent AI labs,” Bank of America analyst Alex Liu wrote in a note cited by CNBC.
The reaction hits similar notes to when the DeepSeek Chinese model jolted Silicon Valley last year. Both models put forward the same argument, stating that China’s open-source approach is closing in on the closed systems US firms are selling at a premium.
“What does it mean for USA to keep its tech advantage?” wrote Xiaoyin Qu, a former Meta product manager, on X. David Sacks, a tech advisor to the Trump administration, called K3’s capabilities “concerning” and cautioned that heavy AI regulation could cost the US ground.
Moonshot’s Kimi K3 arrived just as President Xi Jinping made his first appearance at China’s top AI summit, telling nations that the technology “should not be a solo performance by a single country.”
The U.S. has been moving differently to this sentiment of easy access. OpenAI had just widened public access to GPT-5.6 Sol after the Trump administration cleared it, ending weeks of restricted availability. Anthropic’s most powerful model, Mythos 5, remains limited to a small set of American organizations under a Commerce Department export control order.
Founded in 2023, Moonshot raised $2 billion in May at a valuation above $20 billion, with Alibaba and Tencent among its backers. The real test is set to come on July 27, when the model’s code finally gets released, with the public and developers itching to see if the Kimi K3 model holds up according to reported benchmarks.
The smartest crypto minds already read our newsletter. Want in? Join them.
