メインコンテンツへスキップ

Qwen3 VL 4B (Reasoning)

AlibabaQwenOpen WeightApache 2.0 · Commercial OK

説明

Qwen3-VL is a large multimodal model that unifies vision, language, and reasoning to achieve human-level perception and cognition across text, images, and video. Built on a 235B-parameter architecture, it integrates early joint training of visual and textual modalities for strong language grounding. The model supports up to a 1 million-token context window and excels at visual understanding, spatial reasoning, long video comprehension, and tool-based interaction. It can generate code from images, perform precise 2D/3D object grounding, and operate digital interfaces like a visual agent. The “Instruct” version rivals Gemini 2.5 Pro in perception benchmarks, while the “Thinking” version leads in multimodal reasoning and STEM tasks. With multilingual OCR, creative writing, and fine-grained scene interpretation, Qwen3-VL establishes a new open-source frontier for integrated vision-language intelligence.

リリース日
2025-10-14
パラメータ
4.0B
コンテキスト長
1.0M
モダリティ
image, text

能力レーダー

29
general
17
coding
30
reasoning
29
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 & Tools65
53.0
LS
Code Ranking370
17.0
AA
General Ranking353
32.0
AA
Math Reasoning285
26.0
AA
Multimodal Ranking51
74.0
LS
Reasoning77
54.0
LS
Science378
28.0
AA

ベンチマークスコア (LLM Stats)

3d

BLINK63.4%自己申告

Agents

BFCL-v367.3%自己申告
OSWorld31.4%自己申告

Biology

GPQA64.1%自己申告

Chemistry

SuperGPQA46.8%自己申告

Communication

MM-MT-Bench7.70 / 100自己申告
WritingBench84.0%自己申告
Multi-IF73.6%自己申告

Creativity

Creative Writing v376.1%自己申告
Arena-Hard v236.8%自己申告

Finance

MMLU81.5%自己申告
MMLU-Pro73.6%自己申告
MMLU-ProX65.0%自己申告

General

MMLU-Redux86.0%自己申告
IFEval82.6%自己申告
MLVU-M75.7%自己申告
MMStar73.2%自己申告
MMMU (val)70.8%自己申告
LiveBench 2024112568.4%自己申告
Include64.6%自己申告
MMMU-Pro57.0%自己申告
LiveCodeBench v651.3%自己申告

Grounding

ScreenSpot92.9%自己申告
ScreenSpot Pro49.2%自己申告

Healthcare

VideoMMMU69.4%自己申告

Image To Text

OCRBench80.8%自己申告
OCRBench-V2 (en)61.8%自己申告
OCRBench-V2 (zh)55.8%自己申告

Language

CharadesSTA59.0%自己申告

Long Context

LVBench53.5%自己申告

Math

MathVista-Mini79.5%自己申告
AIME 202574.5%自己申告
MathVision60.0%自己申告
HMMT2553.1%自己申告
PolyMATH44.6%自己申告

Multimodal

DocVQAtest94.2%自己申告
MMBench-V1.186.7%自己申告
AI2D84.9%自己申告
CharXiv-D83.9%自己申告
InfoVQAtest83.0%自己申告
MuirBench75.0%自己申告
CC-OCR73.8%自己申告
MVBench69.3%自己申告
CharXiv-R50.3%自己申告

Reasoning

Hallusion Bench64.1%自己申告
ERQA47.3%自己申告

Spatial Reasoning

RealWorldQA73.2%自己申告

Vision

ODinW39.4%自己申告

AA評価指数

Math Index
25.7
Intelligence Index
13.7
Coding Index
6.7
Mmlu Pro
0.7
Gpqa
0.5
Ifbench
0.4
Livecodebench
0.3
Aime 25
0.3
Lcr
0.2
Scicode
0.2
Tau2
0.2
Hle
0.0
Terminalbench Hard
0.0

LLM Statsカテゴリスコア

Communication
3
Multimodal
100
Structured Output
80
Instruction Following
80
Text-to-image
70
Tool Calling
70
Vision
70
Writing
70
Creativity
70
Finance
70
Grounding
70
Healthcare
70
Image To Text
70
Language
70
Legal
70
Math
70
Reasoning
70
Spatial Reasoning
60
Video
60
3d
60
Biology
60
Chemistry
60
General
60
Physics
60
Agents
50
Economics
50
Long Context
50

価格設定

入力価格無料
出力価格無料
混合価格(3:1)無料

速度

トークン/秒0.0 tokens/s
初トークン遅延0.00s
初回答遅延0.00s

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
DeepInfra100K1.0M

外部リンク