Gemma 4 E4B
Gemma 4 E4B scores 13.7 out of 100 overall, which puts it in a tie for rank 34–36 of 36 models on the AI Vision Index. It is strongest at Not being fooled (39) and weakest at Counting (3). It beats the average of all 36 models in none of the six categories.
Overall score13.795% interval 12.5–15.2
Rank34–36of 36 models (tied)
Cost per 1,000 answers$0 (local)
Seconds per answer0.6 s on a laptop GPU
Thinking effortdefault
Maker Google
WeightsOpen weights
AccessRuns locally
Tested
Reading16
Counting3
Space, 3D and motion10
Spotting changes and errors7
Diagrams7
Not being fooled39
The bar is the score; the light band is the 95% interval. The overall score is the average of the six.
Reading
Reading a clock0
Reading a meter10
Eye chart0
Reading a label54
Reading a chart26
Test strips26
Adding up coins4
Traffic light from a car12
Counting
Counting objects9
Counting stacked goods0
Counting on a conveyor0
Space, 3D and motion
Which is nearer44
Which way it faces8
Same or mirror image0
Will it topple1
Match the render10
Which bay can I park in?5
Follow the hidden object0
Spotting changes and errors
Spot the change12
Shelf against plan10
Rendering glitch5
Check the invoice2
Diagrams
Follow the cable10
Follow the pipe6
Missing dimension4
Not being fooled
Absent objects67
Saying "I can't tell"63
Optical illusions0
Real or AI-made27
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
- MiniCPM-V 4.5 17.1 Compare
- Mistral Small 2603 12.5 Compare
Badge for your site
HTML
<a href="https://aivisionindex.com/models/gemma-4-e4b/"><img src="https://aivisionindex.com/badge/gemma-4-e4b.svg" alt="Gemma 4 E4B: ranked #34–36 of 36 on the AI Vision Index"></a>Markdown
[](https://aivisionindex.com/models/gemma-4-e4b/)Results are CC BY 4.0, so you are free to reuse them if you credit the AI Vision Index. All results (JSON)