Skip to main content

Qwen2.5 VL 72B Instruct

Alibaba Cloud / Qwen TeamQwenOpen Weighttongyi-qianwen

Description

Qwen2.5-VL is the new flagship vision-language model of Qwen, significantly improved from Qwen2-VL. It excels at recognizing objects, analyzing text/charts/layouts in images, acting as a visual agent, understanding long videos (over 1 hour) with event pinpointing, performing visual localization (bounding boxes/points), and generating structured outputs from documents.

Release Date
2025-01-26
Parameters
72.0B
Context Length
32K
Modalities
image, text

Capability Radar

50
general
0
coding
60
reasoning
60
scienceest.
40
agents
80
multimodal

Science uses a reasoning proxy when dedicated science benchmarks are unavailable.

Rankings

Domain#RankScoreSource
Agents & Tools82
45.0
LS
Multimodal Ranking54
73.0
LS
Reasoning74
55.0
LS

Benchmark Scores (LLM Stats)

Agents

AITZ_EM83.2%SR
MobileMiniWob++_SR68.0%SR
AndroidWorld_SR35.0%SR
OSWorld8.8%SR

General

MMVet76.2%SR
MLVU-M74.6%SR
MMStar70.8%SR
MMMU70.2%SR
MMMU-Pro51.1%SR

Grounding

ScreenSpot87.1%SR
ScreenSpot Pro43.6%SR

Image To Text

DocVQA96.4%SR
OCRBench88.5%SR
OCRBench-V2 (en)61.5%SR

Long Context

EgoSchema76.2%SR
LVBench47.3%SR

Math

MathVista-Mini74.8%SR
MathVision38.1%SR

Multimodal

Android Control Low_EM93.7%SR
ChartQA89.5%SR
AI2D88.4%SR
MMBench88.0%SR
CC-OCR79.8%SR
TempCompass74.8%SR
VideoMME w/o sub.73.3%SR
PerceptionTest73.2%SR
MVBench70.4%SR
Android Control High_EM67.4%SR
MMBench-Video2.0%SR

Reasoning

Hallusion Bench55.2%SR

AA Evaluation Indices

No AA evaluation data available

LLM Stats Category Scores

Structured Output
80
Text-to-image
80
Image To Text
80
Spatial Reasoning
70
Grounding
70
Healthcare
70
Reasoning
70
Vision
60
Long Context
60
Math
60
Multimodal
60
Video
50
General
50
Agents
40

Pricing

Input Price$0.25 / 1M tokens
Output Price$0.75 / 1M tokens
Blended Price (3:1)$0.375 / 1M tokens

Speed

No speed data available

Available Providers

(LS internal units)

No provider data available

External Sources