Gemma 3 27b vs.Claude 4 Sonnet
Passed:
Gemma 3 27b 63.0% (51/81)
Claude 4 Sonnet 71.1% (59/83)
Average request time:
Gemma 3 27b 30.87s
Claude 4 Sonnet 19.72s
Passed:
Gemma 3 27b 63.0% (51/81)
Claude 4 Sonnet 71.1% (59/83)
Average request time:
Gemma 3 27b 30.87s
Claude 4 Sonnet 19.72s
Summary
Annotation Understanding
Gemma 3 27b
0 / 2
Claude 4 Sonnet
2 / 2
CAPTCHA
Gemma 3 27b
2 / 2
Claude 4 Sonnet
2 / 2
Color Identification
Gemma 3 27b
0 / 1
Claude 4 Sonnet
0 / 1
Counting nodes
Gemma 3 27b
0 / 0
Claude 4 Sonnet
0 / 1
Defect Detection
Gemma 3 27b
9 / 15
Claude 4 Sonnet
12 / 15
Document Understanding
Gemma 3 27b
8 / 10
Claude 4 Sonnet
9 / 10
Localization
Gemma 3 27b
1 / 1
Claude 4 Sonnet
0 / 1
OCR
Gemma 3 27b
6 / 9
Claude 4 Sonnet
7 / 9
Object Counting
Gemma 3 27b
1 / 9
Claude 4 Sonnet
2 / 10
Object Detection
Gemma 3 27b
2 / 2
Claude 4 Sonnet
2 / 2
Object Measurement
Gemma 3 27b
0 / 2
Claude 4 Sonnet
1 / 2
Object Understanding
Gemma 3 27b
8 / 11
Claude 4 Sonnet
8 / 11
Receipt Reading
Gemma 3 27b
1 / 1
Claude 4 Sonnet
1 / 1
Sign Understanding
Gemma 3 27b
2 / 2
Claude 4 Sonnet
2 / 2
Spatial Relations
Gemma 3 27b
11 / 17
Claude 4 Sonnet
10 / 17
Web Action Understanding
Gemma 3 27b
4 / 4
Claude 4 Sonnet
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!