adjustable glasses illustration

eye chart illustration

clipboard illustration

lenses illustration

ChatGPT-4o vs.Qwen 2.5 VL 7B

Passed:

ChatGPT-4o 78.0% (64/82)

Qwen 2.5 VL 7B 56.1% (46/82)

Average request time:

ChatGPT-4o 13.60s

Qwen 2.5 VL 7B 10.18s

Summary

Annotation Understanding

ChatGPT-4o 1 / 2

Qwen 2.5 VL 7B 2 / 2

CAPTCHA

ChatGPT-4o 2 / 2

Qwen 2.5 VL 7B 2 / 2

Color Identification

ChatGPT-4o 1 / 1

Qwen 2.5 VL 7B 1 / 1

Counting nodes

ChatGPT-4o 0 / 1

Qwen 2.5 VL 7B 0 / 1

Defect Detection

ChatGPT-4o 14 / 15

Qwen 2.5 VL 7B 10 / 15

Document Understanding

ChatGPT-4o 5 / 9

Qwen 2.5 VL 7B 5 / 9

Find an author

ChatGPT-4o 0 / 1

Qwen 2.5 VL 7B 0 / 1

Localization

ChatGPT-4o 1 / 1

Qwen 2.5 VL 7B 0 / 1

OCR

ChatGPT-4o 9 / 9

Qwen 2.5 VL 7B 7 / 9

Object Counting

ChatGPT-4o 2 / 10

Qwen 2.5 VL 7B 1 / 10

Object Detection

ChatGPT-4o 2 / 2

Qwen 2.5 VL 7B 1 / 2

Object Measurement

ChatGPT-4o 1 / 2

Qwen 2.5 VL 7B 0 / 2

Object Understanding

ChatGPT-4o 9 / 10

Qwen 2.5 VL 7B 7 / 10

Receipt Reading

ChatGPT-4o 1 / 1

Qwen 2.5 VL 7B 1 / 1

Sign Understanding

ChatGPT-4o 2 / 2

Qwen 2.5 VL 7B 2 / 2

Spatial Relations

ChatGPT-4o 16 / 17

Qwen 2.5 VL 7B 10 / 17

Web Action Understanding

ChatGPT-4o 4 / 4

Qwen 2.5 VL 7B 3 / 4

Contribute a Prompt

Lightbulb

Have an idea for a prompt? Open a pull request on the project repository!