GPT-5 (high reasoning) vs.Claude 4 Sonnet
Passed:
GPT-5 (high reasoning) 73.5% (61/83)
Claude 4 Sonnet 71.1% (59/83)
Average request time:
GPT-5 (high reasoning) 26.96s
Claude 4 Sonnet 19.72s
Passed:
GPT-5 (high reasoning) 73.5% (61/83)
Claude 4 Sonnet 71.1% (59/83)
Average request time:
GPT-5 (high reasoning) 26.96s
Claude 4 Sonnet 19.72s
Summary
Annotation Understanding
GPT-5 (high reasoning)
1 / 2
Claude 4 Sonnet
2 / 2
CAPTCHA
GPT-5 (high reasoning)
0 / 2
Claude 4 Sonnet
2 / 2
Color Identification
GPT-5 (high reasoning)
1 / 1
Claude 4 Sonnet
0 / 1
Counting nodes
GPT-5 (high reasoning)
0 / 1
Claude 4 Sonnet
0 / 1
Defect Detection
GPT-5 (high reasoning)
12 / 15
Claude 4 Sonnet
12 / 15
Document Understanding
GPT-5 (high reasoning)
8 / 10
Claude 4 Sonnet
9 / 10
Localization
GPT-5 (high reasoning)
1 / 1
Claude 4 Sonnet
0 / 1
OCR
GPT-5 (high reasoning)
9 / 9
Claude 4 Sonnet
7 / 9
Object Counting
GPT-5 (high reasoning)
3 / 10
Claude 4 Sonnet
2 / 10
Object Detection
GPT-5 (high reasoning)
1 / 2
Claude 4 Sonnet
2 / 2
Object Measurement
GPT-5 (high reasoning)
0 / 2
Claude 4 Sonnet
1 / 2
Object Understanding
GPT-5 (high reasoning)
9 / 11
Claude 4 Sonnet
8 / 11
Receipt Reading
GPT-5 (high reasoning)
1 / 1
Claude 4 Sonnet
1 / 1
Sign Understanding
GPT-5 (high reasoning)
2 / 2
Claude 4 Sonnet
2 / 2
Spatial Relations
GPT-5 (high reasoning)
15 / 17
Claude 4 Sonnet
10 / 17
Web Action Understanding
GPT-5 (high reasoning)
4 / 4
Claude 4 Sonnet
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!