Alibabamodel systemcard
Qwen3.8 Max
#12of 27 · general
3.5–3.7 Max previews collapsed
Standing
dot = this model · shaded = the field's middle 60% · track = full fieldGeneral
50.8#12/27
Math
32.6#17/27
Coding
57.0#10/27
Health
37.3#16/26
Engineering
68.8#6/17
Legal
43.1#18/26
Instruct
63.0#7/27
Agents
62.6#9/27
Professional
71.2#6/27
Multimodal
28.7#14/19
Hallucination
39.8#10/27
Speed
21.2#20/22
Value
51.2#10/27
Measured
≈ imputed · amber dot = lab-self-reported among the sourcesGPQA Diamond92.5%#12
HLE42.2%#13
MMLU-Pro89.5%#5
AIME 202581.6%#7
LiveBench73.7%#12
SWE-bench V77.3%#19
LiveCodeBench87.1%#10
SciCode52.9%#12
CritPt13.4%#14
Terminal-Bench69.7%#13
GDPval61.9%#3
Arena1475#10
AA Index58.1#5
Vals Index57.5#15
Epoch Capabilities Index156.1#8
Operations
$2 / $6$ per 1M tokens, in / out
$3blended
51tokens / sec
$0.18$ per solved task
1475arena elo · 3,697 votes