Chinese models match the frontier on nineteen benchmarks

Claim

Alibaba's Qwen3-Max-Thinking now rivals GPT-5.2 and Opus 4.5 across 19 benchmarks and Moonshot's Kimi K2.5 claimed global state of the art on agentic benchmarks, while Grok 4.20 was the only profitable model on PredictionArena.

Reported metric: 19 benchmarks

Reporting date: 2026-01-27 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

• Welcome to January 27, 2026 (theinnermostloop.substack.com)