Qwen3.8 Max
Alibaba · 2 thinking configurations
Intelligence40.2
Input / 1M$2.00
Output / 1M$6.00
Context1M
Tariff: Alibaba · Source: artificial-analysis · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost1 measured configurations · 1 without both values
Selected: thinking max · Click a point to switch configuration
2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
40.2 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
92.7 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
43.0 %SciCodeCode generation for scientific research problems.
53.2 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
81.3 %MMMU-ProMultimodal understanding of academic problems that include images.
82.3 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
31.9 %Specifications from the catalog
Inputtext, image, video, pdf
Outputtext
Tool callingYes
Structured outputYes
Speed38.6 tok/s
First token2.95 s
Maximum output131.1K
Catalog listing date2026-08-03