Qwen3.8-Max
Alibaba · Flagship · Aug 2, 2026
Alibaba's hosted flagship for coding and office work, and the highest-rated model in this catalogue on Arena's vision board. A text-only base checkpoint is published separately.
- Artificial Analysis Intelligence Index
- 45.4#10 of 29
- API price per 1M tokens
- $2 / $6input / output · $3 blended
- Context window
- 1Mtokens
- Output speed
- 37 t/s57s to first answer
Benchmark results
Each row shows this model (blue) against every other model in the catalogue (grey), worst on the left, best on the right.
Overall capability
- Artificial Analysis Intelligence Index45.4#10 of 29
- Epoch Capabilities Index156.4#9 of 16
- LiveBench78.5%#8 of 23
Coding
- Terminal-Bench 4.027%#10 of 13
Claude Code harness, max effort
- Arena WebDev1,672#7 of 27
Preliminary rating
Agents & professional work
- GDPval-AA1,671#7 of 18
Reasoning
- FrontierMath (Tiers 1–3)74.7%#7 of 14
Xhigh effort
- MathArena56.1%#5 of 10
Knowledge & science
- GPQA Diamond92.7%#8 of 14
Xhigh effort
Multimodal
- Arena Vision1,301#1 of 15
Human preference
- Arena Text1,482#9 of 24
Specs & pricing
- Developer
- Alibaba
- Released
- Aug 2, 2026
- Availability
- Generally available
- API model ID
qwen3.8-max- Weights
- Proprietary (API only)
- Parameters
- 2.4T (MoE)
- Reasoning
- Adjustable effort
- Accepts
- Text, Image, Video
- Produces
- Text
- Context window
- 1,000,000 tokens
- Input price
- $2 / 1M tokens
- Output price
- $6 / 1M tokens
Closest alternatives
Models from other labs nearest to Qwen3.8-Max on Artificial Analysis Intelligence Index.
GLM-5.3
Z.ai
Intelligence 44.8 · $1.40 / $4.40
Compare side by sideMiMo-V2.6-Pro
Xiaomi
Intelligence 46.3 · $0.435 / $0.87
Compare side by sideGrok 4.7
SpaceXAI
Intelligence 46.5 · $2 / $6
Compare side by side