Qwen3-VL-8B
Qwen3-VL-8B scores 26.0 out of 100 overall, which puts it at rank 32 of 36 models on the AI Vision Index. It is strongest at Not being fooled (46) and weakest at Counting (18). It beats the average of all 36 models in none of the six categories.
Overall score26.095% interval 24.8–27.6
Rank32of 36 models
Cost per 1,000 answers$0 (local)
Seconds per answer1.3 s on a laptop GPU
Thinking effortdefault
Maker Alibaba
WeightsOpen weights
AccessRuns locally
Tested
Reading33
Counting18
Space, 3D and motion22
Spotting changes and errors18
Diagrams19
Not being fooled46
The bar is the score; the light band is the 95% interval. The overall score is the average of the six.
Reading
Reading a clock41
Reading a meter27
Eye chart0
Reading a label72
Reading a chart46
Test strips30
Adding up coins5
Traffic light from a car41
Counting
Counting objects11
Counting stacked goods18
Counting on a conveyor23
Space, 3D and motion
Which is nearer65
Which way it faces21
Same or mirror image0
Will it topple1
Match the render38
Which bay can I park in?14
Follow the hidden object17
Spotting changes and errors
Spot the change36
Shelf against plan11
Rendering glitch23
Check the invoice4
Diagrams
Follow the cable32
Follow the pipe10
Missing dimension14
Not being fooled
Absent objects90
Saying "I can't tell"86
Optical illusions0
Real or AI-made9
Scores are corrected for guessing: 0 is no better than a guess, 100 is always right. All models on every test
Neighbours on the leaderboard
- Perceptron Mk1.5 28.8 Compare
- MiniCPM-V 4.5 17.1 Compare
Badge for your site
HTML
<a href="https://aivisionindex.com/models/qwen3-vl-8b/"><img src="https://aivisionindex.com/badge/qwen3-vl-8b.svg" alt="Qwen3-VL-8B: ranked #32 of 36 on the AI Vision Index"></a>Markdown
[](https://aivisionindex.com/models/qwen3-vl-8b/)Results are CC BY 4.0, so you are free to reuse them if you credit the AI Vision Index. All results (JSON)