メインコンテンツへスキップ

Qwen3 VL 235B A22B (Reasoning)

AlibabaQwenOpen WeightApache 2.0 · Commercial OK

説明

Qwen3-VL-235B-A22B-Thinking is the most powerful vision-language model in the Qwen series, featuring 236B parameters with MoE architecture for reasoning-enhanced multimodal understanding. Key capabilities include: Visual Agent (operates PC/mobile GUIs, recognizes elements, invokes tools), Visual Coding (generates Draw.io/HTML/CSS/JS from images/videos), Advanced Spatial Perception (2D grounding and 3D grounding for spatial reasoning and embodied AI), Long Context & Video Understanding (native 256K context expandable to 1M, handles hours-long video with second-level indexing), Enhanced Multimodal Reasoning (excels in STEM/Math with causal analysis), Upgraded Visual Recognition (celebrities, anime, products, landmarks, flora/fauna), and Expanded OCR (32 languages, robust in low light/blur/tilt). Architecture innovations include Interleaved-MRoPE for positional embeddings, DeepStack for multi-level ViT feature fusion, and Text-Timestamp Alignment for precise video temporal modeling.

リリース日
2025-09-23
パラメータ
236.0B
コンテキスト長
131K
モダリティ
image, text, video

能力レーダー

42
general
38
coding
86
reasoning
51
science推定
70
agents
100
multimodal

専門的な科学ベンチマークが利用できない場合、Scienceは推論プロキシを使用して推定します。

ランキング

ドメイン#順位スコアソース
Agents & Tools24
66.0
LS
Code Ranking171
47.0
AA
General Ranking146
59.0
AA
Math Reasoning49
89.0
AA
Multimodal Ranking64
67.0
LS
Reasoning37
75.0
LS
Science137
56.0
AA

ベンチマークスコア (LLM Stats)

3d

Objectron0.71 / 100自己申告
BLINK67.1%自己申告
ARKitScenes0.54 / 100自己申告
SUNRGBD0.35 / 100自己申告
Hypersim0.11 / 100自己申告

Agents

SIFO0.77 / 100自己申告
BFCL-v371.9%自己申告
SIFO-Multiturn0.71 / 100自己申告
OSWorld-G0.68 / 100自己申告
OSWorld38.1%自己申告

Chemistry

SuperGPQA64.3%自己申告

Code

Design2Code0.93 / 100自己申告

Communication

MM-MT-Bench8.50 / 100自己申告
WritingBench86.7%自己申告
Multi-IF79.1%自己申告

Creativity

Creative Writing v385.7%自己申告

Embodied

EmbSpatialBench0.84 / 100自己申告
RoboSpatialHome0.74 / 100自己申告

Factuality

SimpleQA44.4%自己申告

Finance

MMLU90.6%自己申告
MMLU-Pro83.8%自己申告
MMLU-ProX80.6%自己申告

General

MMLU-Redux93.7%自己申告
IFEval88.2%自己申告
MMMUval80.6%自己申告
Include80.0%自己申告
LiveBench 2024112579.6%自己申告
MMStar78.7%自己申告
LiveCodeBench v670.1%自己申告
MMMU-Pro69.3%自己申告
SimpleVQA0.61 / 100自己申告

Grounding

ScreenSpot95.4%自己申告
RefCOCO-avg0.92 / 100自己申告
RefSpatialBench0.70 / 100自己申告
ScreenSpot Pro61.8%自己申告

Healthcare

VideoMMMU80.0%自己申告

Image To Text

OCRBench87.5%自己申告
OCRBench-V2 (en)66.8%自己申告
OCRBench-V2 (zh)63.5%自己申告

Instruction Following

MIABench0.93 / 100自己申告

Language

CharadesSTA63.5%自己申告

Long Context

MLVU83.8%自己申告
LVBench63.6%自己申告
MMLongBench-Doc0.56 / 100自己申告

Math

AIME 202589.7%自己申告
MathVista-Mini85.8%自己申告
MathVerse-Mini0.85 / 100自己申告
HMMT2577.4%自己申告
MathVision74.6%自己申告
Humanity's Last Exam13.6%自己申告

Multimodal

DocVQAtest96.5%自己申告
MMBench-V1.190.6%自己申告
InfoVQAtest89.5%自己申告
AI2D89.2%自己申告
CC-OCR81.5%自己申告
MuirBench80.1%自己申告
VideoMME w/o sub.79.0%自己申告
CharXiv-R66.1%自己申告
VisuLogic0.34 / 100自己申告
ZEROBench-Sub0.28 / 100自己申告
ZEROBench0.04 / 100自己申告

Reasoning

ZebraLogic97.3%自己申告
CountBench0.94 / 100自己申告
Hallusion Bench66.7%自己申告
ERQA52.5%自己申告

Spatial Reasoning

RealWorldQA81.3%自己申告

Vision

ODinW43.2%自己申告

AA評価指数

Math Index
88.3
Intelligence Index
27.6
Coding Index
20.9
Aime 25
0.9
Mmlu Pro
0.8
Gpqa
0.8
Livecodebench
0.6
Lcr
0.6
Ifbench
0.6
Tau2
0.5
Scicode
0.4
Terminalbench Hard
0.1
Hle
0.1

LLM Statsカテゴリスコア

Communication
3
Multimodal
100
Writing
90
Creativity
90
Structured Output
80
Text-to-image
80
Video
80
Embodied
80
Finance
80
Grounding
80
Healthcare
80
Instruction Following
80
Language
80
Legal
80
Math
80
Spatial Reasoning
70
Tool Calling
70
Vision
70
General
70
Image To Text
70
Long Context
70
Reasoning
70
Agents
60
Chemistry
60
Economics
60
Physics
60
3d
40
Factuality
40

価格設定

入力価格$0.84 / 1M tokens
出力価格$6.175 / 1M tokens
混合価格(3:1)$2.174 / 1M tokens

速度

トークン/秒30.0 tokens/s
初トークン遅延1.35s
初回答遅延68.10s

利用可能なプロバイダー

(LS内部単位)
プロバイダー入力価格出力価格
DeepInfra450K3.5M
Novita980K4.0M

外部リンク