Story
July 26, 2026
China’s Kimi K3 is squeezing US AI giants from both ends
Chinese AI startup Moonshot has launched its new Kimi K3 model, which reports claim rivals top US models from OpenAI and Anthropic in performance and cost. Following the release, the company temporarily paused new user sign-ups due to overwhelming demand that strained its computing capacity. Moonshot plans to release the model's weights publicly, intensifying the global AI competition.
Moonshot AI’s new Kimi K3 has landed at an awkward moment for the AI industry: just as U.S. labs were trying to defend premium pricing and tight control, a Chinese rival arrived claiming near-frontier performance at lower cost. The result is not just a benchmark fight, but a broader argument over who wins if powerful AI becomes cheaper and more open.
On Moonshot’s account, Kimi K3 is a direct challenge to OpenAI and Anthropic. The company says the model offers “Open Frontier Intelligence” with 2.8 trillion parameters, a 1 million-token context window and native multimodal features.
1 Independent and semi-independent rankings amplified the splash: one widely shared benchmark said Kimi K3 was “now #1 in the Frontend Code Arena,” ahead of Anthropic’s Fable 5.
2 Another assessment found its intelligence “comparable to Opus 4.8 and GPT-5.5” while still trailing Fable 5 and GPT-5.6 Sol overall.3
That has fueled one interpretation in Silicon Valley: China has sharply narrowed the gap. As one prominent repost put it, “the era of the chinese labs being far behind is over.”
4 News coverage has also stressed the strategic difference: Moonshot plans to release Kimi’s weights publicly, making it easier for developers to download, modify and run the model themselves.5
But the backlash has been just as telling. Some U.S. voices argue Kimi’s rise shows America is overregulating itself. David Sacks wrote, “This is concerning,” blaming U.S. policy for “tying itself in knots.”6 He later argued that U.S. “cyber guardrails” were making domestic models less competitive after praising Kimi for handling security-bug fixes others refused.7
Others are trying to cool the panic. A later U.S. assessment shared by Howard Lutnick and amplified by Elon Musk said Kimi K3 “remains behind America’s leading frontier AI models.”
8 That more cautious view matches some reporting: Kimi appears strong enough to pressure prices and business models, even if it has not clearly seized the overall lead.9
One thing is less disputed: demand is real. Moonshot said Kimi K3 had “received far more love than we expected,” and temporarily paused new subscriptions after demand pushed capacity close to its limits.10