Claude 3.7 Sonnet vs.Claude 4 Sonnet
Passed:
Claude 3.7 Sonnet 68.3% (56/82)
Claude 4 Sonnet 70.7% (58/82)
Average request time:
Claude 3.7 Sonnet 16.52s
Claude 4 Sonnet 13.67s
Passed:
Claude 3.7 Sonnet 68.3% (56/82)
Claude 4 Sonnet 70.7% (58/82)
Average request time:
Claude 3.7 Sonnet 16.52s
Claude 4 Sonnet 13.67s
Summary
Annotation Understanding
Claude 3.7 Sonnet
1 / 2
Claude 4 Sonnet
1 / 2
CAPTCHA
Claude 3.7 Sonnet
2 / 2
Claude 4 Sonnet
2 / 2
Color Identification
Claude 3.7 Sonnet
1 / 1
Claude 4 Sonnet
1 / 1
Counting nodes
Claude 3.7 Sonnet
0 / 1
Claude 4 Sonnet
0 / 1
Defect Detection
Claude 3.7 Sonnet
11 / 15
Claude 4 Sonnet
12 / 15
Document Understanding
Claude 3.7 Sonnet
6 / 10
Claude 4 Sonnet
6 / 10
Localization
Claude 3.7 Sonnet
1 / 1
Claude 4 Sonnet
0 / 1
OCR
Claude 3.7 Sonnet
8 / 9
Claude 4 Sonnet
7 / 9
Object Counting
Claude 3.7 Sonnet
3 / 10
Claude 4 Sonnet
3 / 10
Object Detection
Claude 3.7 Sonnet
2 / 2
Claude 4 Sonnet
2 / 2
Object Measurement
Claude 3.7 Sonnet
0 / 2
Claude 4 Sonnet
1 / 2
Object Understanding
Claude 3.7 Sonnet
7 / 10
Claude 4 Sonnet
8 / 10
Receipt Reading
Claude 3.7 Sonnet
1 / 1
Claude 4 Sonnet
1 / 1
Sign Understanding
Claude 3.7 Sonnet
2 / 2
Claude 4 Sonnet
2 / 2
Spatial Relations
Claude 3.7 Sonnet
13 / 17
Claude 4 Sonnet
14 / 17
Web Action Understanding
Claude 3.7 Sonnet
4 / 4
Claude 4 Sonnet
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!