跳轉到主要內容

Qwen3 VL 4B Instruct

AlibabaQwen開源權重Apache 2.0 · 商用許可

描述

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.

發布日期
2025-10-14
參數規模
4.0B
上下文長度
支援模態
image, text

能力雷達圖

22
general
26
coding
37
reasoning
22
science估算
60
agents
90
multimodal

Science 在缺少專門科學評測時使用推理能力代理估算。

排行榜排名

領域#排名分數來源
智慧體能力模型榜94
48.0
LS
程式碼能力榜407
16.0
AA
通用能力榜405
26.0
AA
數學推理237
37.0
AA
多模態榜63
72.0
LS
推理能力95
48.0
LS
科學能力444
20.0
AA

基準測試分數 (LLM Stats)

3d

BLINK65.8%自報

Agents

BFCL-v363.3%自報
OSWorld26.2%自報

Chemistry

SuperGPQA40.3%自報

Communication

MM-MT-Bench7.50 / 100自報
WritingBench82.5%自報

Factuality

SimpleQA48.0%自報

Finance

MMLU77.2%自報
MMLU-Pro67.1%自報
MMLU-ProX59.4%自報

General

IFEval82.3%自報
MMLU-Redux81.5%自報
MLVU-M75.3%自報
MMStar69.8%自報
MMMU (val)67.4%自報
Include61.4%自報
LiveBench 2024112560.9%自報
MMMU-Pro53.2%自報
LiveCodeBench v637.9%自報

Grounding

ScreenSpot94.0%自報
ScreenSpot Pro59.5%自報

Healthcare

VideoMMMU56.2%自報

Image To Text

OCRBench88.1%自報
OCRBench-V2 (en)63.7%自報
OCRBench-V2 (zh)57.6%自報

Language

CharadesSTA55.5%自報

Long Context

LVBench56.2%自報

Math

MathVista-Mini73.7%自報
MathVision51.6%自報
AIME 202546.6%自報
HMMT2530.7%自報
PolyMATH28.8%自報

Multimodal

DocVQAtest95.3%自報
MMBench-V1.185.1%自報
AI2D84.1%自報
InfoVQAtest80.3%自報
CharXiv-D76.2%自報
CC-OCR76.2%自報
MVBench68.9%自報
MuirBench63.8%自報
CharXiv-R39.7%自報

Reasoning

Hallusion Bench57.6%自報
ERQA41.3%自報

Spatial Reasoning

RealWorldQA70.9%自報

Vision

ODinW48.2%自報

AA 評測指數

Math Index
37.0
Intelligence Index
4.1
Mmlu Pro
0.6
Gpqa
0.4
Aime 25
0.4
Ifbench
0.3
Livecodebench
0.3
Tau2
0.2
Scicode
0.1
Lcr
0.1
Hle
0.0
Terminalbench Hard
0.0

LLM Stats 分類評分

Communication
4
Multimodal
90
Instruction Following
80
Grounding
80
Creativity
80
Text-to-image
80
Writing
80
Image To Text
70
Language
70
Legal
70
Structured Output
70
3d
70
Long Context
60
Math
60
Reasoning
60
Spatial Reasoning
60
Finance
60
General
60
Healthcare
60
Tool Calling
60
Video
60
Vision
60
Factuality
50
Physics
40
Agents
40
Chemistry
40
Economics
40

定價

輸入價格免費
輸出價格免費
混合價格(3:1)免費

速度

Tokens/秒0.0
首Token延遲0.00s
首回答延遲0.00s

供應商價格排行

供應商價格排行

1 個供應商

供應商輸入輸出
1DeepInfra
$0
$0

比較該模型在不同 API 供應商之間的定價。

外部連結