adjustable glasses illustration

eye chart illustration

clipboard illustration

lenses illustration

Gemini 3 Flash vs.Gemma 4 31B

Passed:

Gemini 3 Flash 73.2% (60/82)

Gemma 4 31B 72.2% (39/54)

Average request time:

Gemini 3 Flash 8.50s

Gemma 4 31B 15.30s

Summary

Annotation Understanding

Gemini 3 Flash 1 / 2

Gemma 4 31B 0 / 1

CAPTCHA

Gemini 3 Flash 2 / 2

Gemma 4 31B 2 / 2

Color Identification

Gemini 3 Flash 0 / 1

Gemma 4 31B 0 / 0

Defect Detection

Gemini 3 Flash 11 / 15

Gemma 4 31B 6 / 8

Document Understanding

Gemini 3 Flash 9 / 10

Gemma 4 31B 7 / 7

Localization

Gemini 3 Flash 1 / 1

Gemma 4 31B 0 / 1

OCR

Gemini 3 Flash 9 / 9

Gemma 4 31B 4 / 4

Object Counting

Gemini 3 Flash 3 / 11

Gemma 4 31B 2 / 8

Object Detection

Gemini 3 Flash 2 / 2

Gemma 4 31B 1 / 2

Object Measurement

Gemini 3 Flash 0 / 1

Gemma 4 31B 0 / 0

Object Understanding

Gemini 3 Flash 8 / 11

Gemma 4 31B 4 / 7

Receipt Reading

Gemini 3 Flash 1 / 1

Gemma 4 31B 1 / 1

Sign Understanding

Gemini 3 Flash 2 / 2

Gemma 4 31B 1 / 1

Spatial Relations

Gemini 3 Flash 13 / 17

Gemma 4 31B 10 / 13

Web Action Understanding

Gemini 3 Flash 4 / 4

Gemma 4 31B 4 / 4

Contribute a Prompt

Lightbulb

Have an idea for a prompt? Open a pull request on the project repository!