adjustable glasses illustration eye chart illustration clipboard illustration lenses illustration

GPT-5.4 vs.Qwen 3.5 2B

Passed:
GPT-5.4 79.3% (65/82)
Qwen 3.5 2B 47.6% (39/82)
Average request time:
GPT-5.4 9.40s
Qwen 3.5 2B 6.33s

Summary

Annotation Understanding
GPT-5.4 2 / 2
Qwen 3.5 2B 0 / 2
CAPTCHA
GPT-5.4 0 / 2
Qwen 3.5 2B 0 / 2
Color Identification
GPT-5.4 1 / 1
Qwen 3.5 2B 0 / 1
Defect Detection
GPT-5.4 13 / 15
Qwen 3.5 2B 6 / 15
Document Understanding
GPT-5.4 9 / 10
Qwen 3.5 2B 7 / 10
Localization
GPT-5.4 1 / 1
Qwen 3.5 2B 0 / 1
OCR
GPT-5.4 9 / 9
Qwen 3.5 2B 6 / 9
Object Counting
GPT-5.4 4 / 11
Qwen 3.5 2B 3 / 11
Object Detection
GPT-5.4 2 / 2
Qwen 3.5 2B 1 / 2
Object Measurement
GPT-5.4 0 / 1
Qwen 3.5 2B 0 / 1
Object Understanding
GPT-5.4 8 / 11
Qwen 3.5 2B 3 / 11
Receipt Reading
GPT-5.4 1 / 1
Qwen 3.5 2B 1 / 1
Sign Understanding
GPT-5.4 2 / 2
Qwen 3.5 2B 2 / 2
Spatial Relations
GPT-5.4 15 / 17
Qwen 3.5 2B 8 / 17
Web Action Understanding
GPT-5.4 4 / 4
Qwen 3.5 2B 4 / 4

Contribute a Prompt