What smart people are saying about China's hot new Kimi K3 AI model

• Chinese startup Moonshot AI released Kimi K3, a powerful new model that has the tech world talking.
• Kimi K3's release highlights China's narrowing of the gap with the US on AI.
• Leaders from the US tech sector have been reacting to the new model online.

Chinese AI startup Moonshot AI released Kimi K3 on Thursday, a powerful new model that the startup says is the largest open-weight AI system in the world.

Moonshot says K3 still trails the most powerful proprietary models, Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol, overall, but beats the labs' second-tier systems, Claude Opus 4.8 and GPT-5.5, on benchmarks including coding and agentic tasks.

Within a day, K3 topped Arena's frontend coding leaderboard, ahead of every leading US model, and placed third on Artificial Analysis's Intelligence Index.

The release, timed just ahead of the 2026 World Artificial Intelligence Conference in Shanghai, is the latest sign that Chinese labs are closing the gap with leading US systems.

It also rattled AI-related markets: Moonshot, backed by Alibaba and Tencent, is valued at roughly $31.5 billion, a fraction of the trillion-dollar-plus valuations attached to Anthropic and OpenAI.

Here's what smart people in the worlds of tech and academia are saying about it.

David Sacks, the venture capitalist who served as President Donald Trump's first AI and crypto czar before moving in March to cochair the President's Council of Advisors on Science and Technology, called the release "concerning."

In a Friday X post sharing the Arena leaderboard, Sacks noted it was the first time a Chinese model had taken the top spot for frontend coding, with Kimi K3 also scoring at or near the frontier on other benchmarks.

Sacks argued the US is hobbling itself in response: blocking new data centers, layering on state regulations, and pushing for federal agencies to pre-approve frontier models. "This is how you lose the AI race," he wrote, warning that the rest of the world won't play by America's rules if it bogs itself down.

"Permissionless innovation" is how America won the internet, Sacks said, adding that the US can win in AI while addressing risks in a targeted way, "or we'll watch our lead evaporate."

Aaron Levie said the release was a "huge win" for companies building on AI.

In an X post on Thursday, the Box CEO congratulated the Kimi team and said it was "truly wild" to see this level of performance from open models, pointing to Kimi K3's third-place ranking on Artificial Analysis' Intelligence Index — behind only Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol.

Levie argued that cheaper frontier-level intelligence directly expands what enterprises can do with AI. There's a large backlog of workflows companies would love to automate, he said, held back only by token costs.

Ethan Mollick offered "a note of caution" amid the hype.

The Wharton professor, who studies AI's effects on work, took to X to say Kimi K3 "messed up in a bunch of ways" when he asked it to perform a complex statistical audit of some of his prior academic work, including misapplying statistical methods.

Mollick shared a detailed critique of K3's audit — generated, notably, by OpenAI's rival GPT-5.6 Pro model — that identified errors in the audit's core statistical approach. Mollick said he agreed with the critique.

Venture capitalist and "All-In Podcast" cohost Jason Calacanis said the pace of AI progress is accelerating, and made some bold predictions.

"It's happening folks," Calacanis wrote on X, sharing an Arena leaderboard showing Kimi K3 ranked first for frontend coding, ahead of Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol.

He argued the field has moved faster in the last 30 days, across a dozen players, than in the previous year, with open-source models "compounding" while frontier labs refine.

Calacanis predicted things will "get wild" when open-source AI reaches robotics, self-driving, and life sciences, and said that 2026 will be the year of AGI, with superintelligence following in 2027 or 2028.

"ITS GONNA GET VERY STRANGE," he wrote.

Russ Salakhutdinov, the CMU professor who co-advised Moonshot founder and CEO Zhilin Yang's Ph.D., called the release "a huge win for the open-source community."

In an X post, Salakhutdinov congratulated his former student, resurfacing a 2019 post celebrating Yang's thesis defense.

Yang completed his Ph.D. at Carnegie Mellon in just four years, during which he contributed influential research, including Transformer-XL and XLNet, which helped shape the architecture of modern language models.

"It feels like just yesterday Zhilin was graduating from my lab at CMU," Salakhutdinov wrote, thanking Yang and the Kimi team "for everything you're doing for the open-source community."

Gary Marcus, a longtime critic of the AI industry's economics, had a blunter reaction.

"Congress should investigate. Seriously," Marcus wrote on X, quoting a Goldman Sachs chart showing capital expenditure for major US cloud providers projected to reach roughly $1 trillion in 2027, about eight times the projected spending of their Chinese counterparts.