The Agent RaceJuly 16, 2026via The Decoder

Kimi's open model K3 nears GPT-5.6 Sol and Fable 5 while signaling the end of super cheap Chinese AI

Why it matters

Kimi's K3 open-weight model demonstrates that cutting-edge capability is no longer the exclusive domain of US labs, while signaling a market shift: competitive Chinese models are moving upmarket, abandoning the race-to-the-bottom pricing that characterized the previous generation.

Key signals

  • K3: 2.8 trillion parameters, 1M context window
  • Benchmark performance: competitive with Claude Fable 5 and GPT-5.6 Sol
  • Beats Opus 4.8 and GLM 5.2 on some benchmarks
  • Full weights release scheduled July 27
  • K3 pricing significantly higher than predecessor
  • Signals end of ultra-cheap Chinese AI commoditization

The hook

2.8 trillion parameters. Kimi's K3 just closed the gap with GPT-5.6 Sol—and killed the era of $0.01 Chinese AI.

Kimi is launching K3, a multimodal open-weight model with 2.8 trillion parameters and one million tokens of context. In the company's own benchmarks, it comes close to Claude Fable 5 and GPT 5.6 Sol while beating Opus 4.8 and GLM 5.2, in some cases by a wide margin. The model is also significantly pricier than its predecessor. Full weights are scheduled for release by July 27.

The week's key stories, every Friday.

For practitioners and enthusiasts — free, in your inbox.

Free forever. No spam.