adjustable glasses illustration

eye chart illustration

clipboard illustration

lenses illustration

Qwen 3.5 4B vs.Phi 4 Multimodal

Passed:

Qwen 3.5 4B 63.4% (52/82)

Phi 4 Multimodal 49.4% (41/83)

Average request time:

Qwen 3.5 4B 5.81s

Phi 4 Multimodal 15.92s

Summary

Annotation Understanding

Qwen 3.5 4B 0 / 2

Phi 4 Multimodal 1 / 2

CAPTCHA

Qwen 3.5 4B 2 / 2

Phi 4 Multimodal 0 / 2

Color Identification

Qwen 3.5 4B 0 / 1

Phi 4 Multimodal 0 / 1

Defect Detection

Qwen 3.5 4B 12 / 15

Phi 4 Multimodal 11 / 15

Document Understanding

Qwen 3.5 4B 7 / 10

Phi 4 Multimodal 7 / 10

Localization

Qwen 3.5 4B 0 / 1

Phi 4 Multimodal 0 / 1

OCR

Qwen 3.5 4B 7 / 9

Phi 4 Multimodal 5 / 9

Object Counting

Qwen 3.5 4B 4 / 11

Phi 4 Multimodal 0 / 11

Object Detection

Qwen 3.5 4B 2 / 2

Phi 4 Multimodal 2 / 2

Object Measurement

Qwen 3.5 4B 0 / 1

Phi 4 Multimodal 0 / 1

Object Understanding

Qwen 3.5 4B 7 / 11

Phi 4 Multimodal 5 / 11

Receipt Reading

Qwen 3.5 4B 1 / 1

Phi 4 Multimodal 1 / 1

Sign Understanding

Qwen 3.5 4B 2 / 2

Phi 4 Multimodal 2 / 2

Spatial Relations

Qwen 3.5 4B 8 / 17

Phi 4 Multimodal 7 / 17

Web Action Understanding

Qwen 3.5 4B 4 / 4

Phi 4 Multimodal 2 / 4

Contribute a Prompt

Lightbulb

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