GPT-4.1 vs.Claude 3.5 Haiku
Passed:
GPT-4.1 72.3% (60/83)
Claude 3.5 Haiku 65.1% (54/83)
Average request time:
GPT-4.1 9.74s
Claude 3.5 Haiku 16.75s
Passed:
GPT-4.1 72.3% (60/83)
Claude 3.5 Haiku 65.1% (54/83)
Average request time:
GPT-4.1 9.74s
Claude 3.5 Haiku 16.75s
Summary
Annotation Understanding
GPT-4.1
1 / 2
Claude 3.5 Haiku
2 / 2
CAPTCHA
GPT-4.1
2 / 2
Claude 3.5 Haiku
0 / 2
Color Identification
GPT-4.1
0 / 1
Claude 3.5 Haiku
0 / 1
Counting nodes
GPT-4.1
0 / 1
Claude 3.5 Haiku
0 / 1
Defect Detection
GPT-4.1
12 / 15
Claude 3.5 Haiku
10 / 15
Document Understanding
GPT-4.1
8 / 10
Claude 3.5 Haiku
8 / 10
Localization
GPT-4.1
1 / 1
Claude 3.5 Haiku
0 / 1
OCR
GPT-4.1
9 / 9
Claude 3.5 Haiku
7 / 9
Object Counting
GPT-4.1
2 / 10
Claude 3.5 Haiku
5 / 10
Object Detection
GPT-4.1
2 / 2
Claude 3.5 Haiku
2 / 2
Object Measurement
GPT-4.1
0 / 2
Claude 3.5 Haiku
0 / 2
Object Understanding
GPT-4.1
8 / 11
Claude 3.5 Haiku
7 / 11
Receipt Reading
GPT-4.1
1 / 1
Claude 3.5 Haiku
1 / 1
Sign Understanding
GPT-4.1
2 / 2
Claude 3.5 Haiku
2 / 2
Spatial Relations
GPT-4.1
14 / 17
Claude 3.5 Haiku
8 / 17
Web Action Understanding
GPT-4.1
4 / 4
Claude 3.5 Haiku
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!