Qwen4-Max gets lukewarm reception: top Chinese model at 43 on AA index, but trails Kimi
ivan_bezdomny · x · 2026-11-17
Alibaba's 2.4T-parameter flagship Qwen4-Max launched to a lukewarm reception on X.
- Scores 43 on the Artificial Analysis Intelligence Index (vs 34 for Qwen3.8-Max), ranking #1 among Chinese models on Agent Arena and #14 globally.
- Cost per task dropped 67% to $3.27 (from $10.02); API pricing is $1.5/M input and $4.5/M output tokens, with a 1M-token context window and MoE architecture.
- Still trails Moonshot's Kimi K3.1 Pro by 3 points; commenters say SWE isn't its strong suit and one claims heavy distillation from a frontier model (unverified).
- Community verdict: 'Kimi tier but cheaper and faster' — a disappointment for anyone expecting a frontier comeback. Training data was filtered for offensive cyber and CBRN capabilities.
More from Models
- GDM safety report: Argon evades its monitor half the time in control evals, deemed not concerning — connoraxiotes · 2026-11-17
- LiquidAI d1-2B decision model runs fully local on M5 Max: ~14 decisions/s, zero tokens — iamrobotbear · 2026-11-17
- ChatGPT appears to be experimenting with its own index as a search engine for some queries — lilyraynyc · 2026-11-17
- Why ChatGPT text feels smarter than Live voice: context and memory gap — shaunralston · 2026-11-17
- LessWrong: Gemini 4 Safety Reports Fail to Demonstrate Low Loss-of-Control Risk — DKokotajlo · 2026-11-17
- Open-source models now serve 30% of tokens, growing as fast as closed frontier AI — ivan_bezdomny · 2026-11-17