Compare Start here

Alibaba QwenLanguage modelSuperseded

Qwen3-Max

Trillion-parameter Qwen that brought Alibaba into the top tier.

Compare head-to-head Race it in 3D

Overall Intelligence

#43

of 56
19.9 pts behind #1

0

races led

8

weeks at #1

68.8

score

Released: Sep 24, 2025; 1.1 yr ago
Context: 256K; 64K max output
Price: $1.2 / $6; per 1M in/out · $2.40 blended
Speed: 60 tok/s; 1.5s to first token
Modalities: text; outputs text
Weights: Closed; API and apps

The 30-second version

Trillion-parameter Qwen that brought Alibaba into the top tier.

Key takeaways

1Trillion-parameter Qwen that brought Alibaba into the top tier.

Where it ranks

Current snapshot · illustrative scores · click any race to watch it

Qwen3-MaxBest on each axis

Overall

43Overall+19.9 pts

Coding

41Coding+19.9 pts 42Frontend & UI+130 Elo 38Backend & APIs+16.7 pts 40Distributed Systems+20.3 pts 39Full-Stack Apps+18.4 pts 41Debugging & Repair+20.1 pts 39Agentic Coding+21.2 pts 40Code Review & Security+17.7 pts 43Data Science & SQL+19.6 pts 39Mobile Apps+17.0 pts 42Terminal & DevOps+24.4 pts 39Competitive Programming+527

Hard Reasoning

41Hard Reasoning+20.1 pts 36Frontier Questions+30.7 pts 44Abstract & Novel Reasoning+33.9 pts 39Logic & Puzzles+16.2 pts 43Planning & Strategy+19.1 pts

Math

38Math+18.4 pts 37Competition Math+14.0 pts 39Research Math+24.0 pts 37Applied & Quantitative+15.4 pts

Science

41Science+18.8 pts 42Graduate Science QA+12.2 pts 42Biology & Medicine+17.2 pts 39Physics & Chemistry+15.1 pts 38Research Assistance+16.5 pts

Writing & Language

37Writing & Language+17.2 pts 38Creative Writing+114 Elo 37Technical Writing+13.7 pts 37Editing & Style+13.7 pts 30Multilingual+9.4 pts 39Instruction Following+13.3 pts 39Conversation & EQ+108 Elo

Multimodal

49Multimodal+56.1 pts 49Image Understanding+41.0 pts 48Documents & Charts+39.9 pts 48Video Understanding+55.9 pts 49Speech & Audio+56.8 pts

Tool Use & Agents

44Tool Use & Agents+22.8 pts 39Tool Use+15.1 pts 46Computer Use+43.2 pts 41Web Research+23.2 pts 40Long-Horizon+26.0 hrs

Long Context

41Long Context+22.9 pts 391M Recall+17.6 pts 43Long Reasoning+21.9 pts

Speed & Cost

25Speed & Cost+39.2 pts 44Output Speed+462 tok/s 15Latency+1.06s 37Cost-Efficiency+50.0 pts

Trust & Factuality

40Trust & Factuality+18.9 pts 40Factuality+15.6 pts 38Honesty & Calibration+15.4 pts

Overall Intelligence over time

#1 frontierQwen3-MaxV4.1 FlashGLM-5.3

Update history

The distilled version first; open any release for the deeper breakdown.

The Qwen Max family

Qwen3April 2025 Qwen3-MaxSeptember 2025 Qwen3.6-MaxApril 2026