Moonshot's Kimi K3 jolts AI stocks before its weights are public

Yang Zhilin's 2.8 trillion-parameter model topped a coding leaderboard, while Moonshot concedes it still trails the leading US systems overall.

By ยท Published

Why it matters

K3 shows how a narrow but credible benchmark win can challenge the economics of closed AI models before outsiders can inspect the underlying weights or training costs.

abstract symbolic representation of the story's core idea (editorial illustration in the spirit of New Yorker or The Atlantic)

Yang Zhilin (@lostsm), Moonshot AI's co-founder and chief executive, turned a model release into a market event on July 16th when Kimi K3's coding results challenged the assumption that US laboratories held a durable technical lead.

AI and semiconductor shares fell the following day as investors compared K3 with DeepSeek's market-shaking arrival in 2025, Bloomberg reported. The reaction arrived before developers could inspect K3's full weights, which Moonshot AI says it will publish by July 27th.

Reader comments

Conversation for this story loads after sign-in.