Updated
Updated · CNBC · Jul 17
Moonshot AI Debuts 2.8 Trillion-Parameter Kimi K3 as Chinese Rivals’ Shares Slide
Updated
Updated · CNBC · Jul 17

Moonshot AI Debuts 2.8 Trillion-Parameter Kimi K3 as Chinese Rivals’ Shares Slide

3 articles · Updated · CNBC · Jul 17

Summary

  • Moonshot AI said Friday its new Kimi K3 model outperformed OpenAI’s GPT 5.5 and Anthropic’s Claude Opus 4.8 on coding and agent benchmarks, though it still trails GPT 5.6 Sol and Claude Fable 5 overall.
  • The 2.8 trillion-parameter model is China’s largest so far, and Bank of America said it shows Chinese labs can still post major gains despite domestic hardware and compute constraints.
  • Chinese AI stocks fell after the release, with Z.ai plunging 28%, MiniMax dropping 16% and Alibaba slipping 4% as analysts said K3 raises the bar for independent model builders.
  • The launch adds pressure in the U.S.-China AI race as Chinese models gain Western users by narrowing performance gaps while staying cheaper than top American systems, drawing scrutiny from U.S. lawmakers.

Insights

As China's Kimi K3 challenges top US models, is the high-cost, closed-source AI business model now obsolete?
Firms flee to open-source AI for data privacy, but can we trust models allegedly trained on stolen data?
With AI valuations hitting bubble territory, can any company justify a $31.5B price tag in a commoditizing market?