Qwen3.5 122B A10B
Alibaba · 2 thinking configurations
Intelligence15.6
Input / 1M$0.40
Output / 1M$3.20
Context262.1K
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 thinking · Click a point to switch configuration
The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
15.6 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
85.7 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
25.2 %SciCodeCode generation for scientific research problems.
39.7 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
47.6 %τ²-BenchTool use and interaction in assistance scenarios. Does not replace an evaluation of your integration.
93.6 %IFBenchVerifiable instruction following. A proxy for precision rather than a benchmark of creative style.
75.7 %MMMU-ProMultimodal understanding of academic problems that include images.
75.0 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
24.4 %Specifications from the catalog
Inputtext, image, video
Outputtext
Tool callingYes
Structured outputYes
Speed124.5 tok/s
First token2.39 s
Maximum output65.5K
Catalog listing date2026-02-25