adjustable glasses illustration eye chart illustration clipboard illustration lenses illustration

Claude 3.5 Haiku vs.GPT-5 (high reasoning)

Passed:
Claude 3.5 Haiku 65.1% (54/83)
GPT-5 (high reasoning) 73.5% (61/83)
Average request time:
Claude 3.5 Haiku 16.75s
GPT-5 (high reasoning) 26.96s

Summary

Annotation Understanding
Claude 3.5 Haiku 2 / 2
GPT-5 (high reasoning) 1 / 2
CAPTCHA
Claude 3.5 Haiku 0 / 2
GPT-5 (high reasoning) 0 / 2
Color Identification
Claude 3.5 Haiku 0 / 1
GPT-5 (high reasoning) 1 / 1
Counting nodes
Claude 3.5 Haiku 0 / 1
GPT-5 (high reasoning) 0 / 1
Defect Detection
Claude 3.5 Haiku 10 / 15
GPT-5 (high reasoning) 12 / 15
Document Understanding
Claude 3.5 Haiku 8 / 10
GPT-5 (high reasoning) 8 / 10
Localization
Claude 3.5 Haiku 0 / 1
GPT-5 (high reasoning) 1 / 1
OCR
Claude 3.5 Haiku 7 / 9
GPT-5 (high reasoning) 9 / 9
Object Counting
Claude 3.5 Haiku 5 / 10
GPT-5 (high reasoning) 3 / 10
Object Detection
Claude 3.5 Haiku 2 / 2
GPT-5 (high reasoning) 1 / 2
Object Measurement
Claude 3.5 Haiku 0 / 2
GPT-5 (high reasoning) 0 / 2
Object Understanding
Claude 3.5 Haiku 7 / 11
GPT-5 (high reasoning) 9 / 11
Receipt Reading
Claude 3.5 Haiku 1 / 1
GPT-5 (high reasoning) 1 / 1
Sign Understanding
Claude 3.5 Haiku 2 / 2
GPT-5 (high reasoning) 2 / 2
Spatial Relations
Claude 3.5 Haiku 8 / 17
GPT-5 (high reasoning) 15 / 17
Web Action Understanding
Claude 3.5 Haiku 4 / 4
GPT-5 (high reasoning) 4 / 4

Contribute a Prompt