Kimi K3 Ranks #1 in AI Benchmarks Ahead of US Models

Full model weights won’t arrive until July 27, but leading benchmark portal Arena AI currently ranks Kimi K3 at #1 on its Frontend Code competition, not just above but considerably ahead of Anthropic’s Fable 5 and Opus 4.8, OpenAI’s GPT-5.6 Sol, and SpaceXAI’s Grok-4.5. Kimi’s ranked #1 in 6 out of 7 domains, including Brand & Marketing, Data & Analytics, Consumer Product, and Content Creation Tools. It drops to second place, behind Fable 5, in just one category: Gaming. On X, former AI czar (and friend of the pod) David Sacks called the development “concerning,” and cautioned that America risks losing the AI race entirely as states move to further regulate model releases and ban data center projects. National security aside, Kimi K3 also poses a potential threat to massive American closed-source frontier labs like Anthropic and OpenAI. How can they compete, long-term, if Chinese open source labs can replicate their results within months, if not weeks, at considerably lower costs? #🇨🇳 ChinA.I. 🤖🧠🦾🤖 The Kimi release was timed to coincide with China’s World AI Conference in Shanghai this week, which received its first appearance from the nation’s president, Xi Jinping. He called for “international cooperation” on #Artificial Intelligence innovation and safety, noting that these important decisions should not be left up to any single country. https://x.com/DavidSacks/..