GPT-5.4 vs.Claude 4.6 Opus
Passed:
GPT-5.4 79.3% (65/82)
Claude 4.6 Opus 70.7% (58/82)
Average request time:
GPT-5.4 9.40s
Claude 4.6 Opus 12.68s
Passed:
GPT-5.4 79.3% (65/82)
Claude 4.6 Opus 70.7% (58/82)
Average request time:
GPT-5.4 9.40s
Claude 4.6 Opus 12.68s
Summary
Annotation Understanding
GPT-5.4
2 / 2
Claude 4.6 Opus
0 / 2
CAPTCHA
GPT-5.4
0 / 2
Claude 4.6 Opus
2 / 2
Color Identification
GPT-5.4
1 / 1
Claude 4.6 Opus
0 / 1
Defect Detection
GPT-5.4
13 / 15
Claude 4.6 Opus
10 / 15
Document Understanding
GPT-5.4
9 / 10
Claude 4.6 Opus
8 / 10
Localization
GPT-5.4
1 / 1
Claude 4.6 Opus
1 / 1
OCR
GPT-5.4
9 / 9
Claude 4.6 Opus
9 / 9
Object Counting
GPT-5.4
4 / 11
Claude 4.6 Opus
3 / 11
Object Detection
GPT-5.4
2 / 2
Claude 4.6 Opus
2 / 2
Object Measurement
GPT-5.4
0 / 1
Claude 4.6 Opus
1 / 1
Object Understanding
GPT-5.4
8 / 11
Claude 4.6 Opus
8 / 11
Receipt Reading
GPT-5.4
1 / 1
Claude 4.6 Opus
1 / 1
Sign Understanding
GPT-5.4
2 / 2
Claude 4.6 Opus
2 / 2
Spatial Relations
GPT-5.4
15 / 17
Claude 4.6 Opus
11 / 17
Web Action Understanding
GPT-5.4
4 / 4
Claude 4.6 Opus
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!