#13
Qwen 3.8 Max
qwen3.8-max
0.0
Helpfulness
Instruction Following
Comprehension
Empathy
Creative Writing
Helpfulness
0.0
Instruction Following
0.0
Comprehension
0.0
Empathy
0.0
Creative Writing
0.0
Speed
Avg 52 tok/s
Overall Assistant Score
An average score combining the 5 main categories.
91.70 pts
Rank #13
88th Percentile
0
Novice
33
Capable
66
Proficient
100
Expert
Qwen 3.8 Max is Alibaba's August 3rd, 2026 hosted flagship at $2/$6. Our tests score helpfulness 94.5, instruction following 95.5, and comprehension 95.5 — tied with the top open assistants for getting work done. Empathy is 86.5: capable and polite, cooler than Claude. Creative writing is 86.5: clean and versatile, less literary than Claude Opus. Speed is 52.0 with heavier reasoning. A strong general assistant when you want GLM-class follow-through at a flat hosted price.

Overall Score vs Price

Price axis is logarithmic (USD per 1M input tokens) · Higher score and lower price is better

Ringed point is this model

Anthropic
Baidu
Bytedance
DeepSeek
Google
Meta
Minimax
Mistral
Moonshot
OpenAI
Qwen
SpaceX AI
Tencent
Xiaomi
Zhipu AI

Compare this model

Intelligence

Overall Score · Higher is better

chatio

The benchmark for real-world helpfulness. Evaluating LLMs on practical, everyday tasks that provide a clear picture of their real-world capabilities.

Designed and developed by Alex Tyh Wang

© Chatio 2026. All rights reserved.