Qwen 3.5 0.8B vs.SmolVLM2 2.2B
Passed:
Qwen 3.5 0.8B 32.9% (27/82)
SmolVLM2 2.2B 36.6% (30/82)
Average request time:
Qwen 3.5 0.8B 4.36s
SmolVLM2 2.2B 4.16s
Passed:
Qwen 3.5 0.8B 32.9% (27/82)
SmolVLM2 2.2B 36.6% (30/82)
Average request time:
Qwen 3.5 0.8B 4.36s
SmolVLM2 2.2B 4.16s
Summary
Annotation Understanding
Qwen 3.5 0.8B
0 / 2
SmolVLM2 2.2B
2 / 2
CAPTCHA
Qwen 3.5 0.8B
0 / 2
SmolVLM2 2.2B
0 / 2
Color Identification
Qwen 3.5 0.8B
0 / 1
SmolVLM2 2.2B
0 / 1
Defect Detection
Qwen 3.5 0.8B
7 / 15
SmolVLM2 2.2B
9 / 15
Document Understanding
Qwen 3.5 0.8B
4 / 10
SmolVLM2 2.2B
3 / 10
Localization
Qwen 3.5 0.8B
0 / 1
SmolVLM2 2.2B
0 / 1
OCR
Qwen 3.5 0.8B
6 / 9
SmolVLM2 2.2B
2 / 9
Object Counting
Qwen 3.5 0.8B
1 / 11
SmolVLM2 2.2B
1 / 11
Object Detection
Qwen 3.5 0.8B
1 / 2
SmolVLM2 2.2B
1 / 2
Object Measurement
Qwen 3.5 0.8B
0 / 1
SmolVLM2 2.2B
0 / 1
Object Understanding
Qwen 3.5 0.8B
2 / 11
SmolVLM2 2.2B
3 / 11
Receipt Reading
Qwen 3.5 0.8B
0 / 1
SmolVLM2 2.2B
0 / 1
Sign Understanding
Qwen 3.5 0.8B
1 / 2
SmolVLM2 2.2B
2 / 2
Spatial Relations
Qwen 3.5 0.8B
4 / 17
SmolVLM2 2.2B
10 / 17
Web Action Understanding
Qwen 3.5 0.8B
2 / 4
SmolVLM2 2.2B
0 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!