Gemma 4 26B (A4B) vs.Gemma 4 31B
Passed:
Gemma 4 26B (A4B) 62.2% (51/82)
Gemma 4 31B 72.2% (39/54)
Average request time:
Gemma 4 26B (A4B) 8.84s
Gemma 4 31B 15.30s
Passed:
Gemma 4 26B (A4B) 62.2% (51/82)
Gemma 4 31B 72.2% (39/54)
Average request time:
Gemma 4 26B (A4B) 8.84s
Gemma 4 31B 15.30s
Summary
Annotation Understanding
Gemma 4 26B (A4B)
1 / 2
Gemma 4 31B
0 / 1
CAPTCHA
Gemma 4 26B (A4B)
2 / 2
Gemma 4 31B
2 / 2
Color Identification
Gemma 4 26B (A4B)
1 / 1
Gemma 4 31B
0 / 0
Defect Detection
Gemma 4 26B (A4B)
8 / 15
Gemma 4 31B
6 / 8
Document Understanding
Gemma 4 26B (A4B)
7 / 10
Gemma 4 31B
7 / 7
Localization
Gemma 4 26B (A4B)
1 / 1
Gemma 4 31B
0 / 1
OCR
Gemma 4 26B (A4B)
8 / 9
Gemma 4 31B
4 / 4
Object Counting
Gemma 4 26B (A4B)
2 / 11
Gemma 4 31B
2 / 8
Object Detection
Gemma 4 26B (A4B)
1 / 2
Gemma 4 31B
1 / 2
Object Measurement
Gemma 4 26B (A4B)
0 / 1
Gemma 4 31B
0 / 0
Object Understanding
Gemma 4 26B (A4B)
7 / 11
Gemma 4 31B
4 / 7
Receipt Reading
Gemma 4 26B (A4B)
1 / 1
Gemma 4 31B
1 / 1
Sign Understanding
Gemma 4 26B (A4B)
1 / 2
Gemma 4 31B
1 / 1
Spatial Relations
Gemma 4 26B (A4B)
11 / 17
Gemma 4 31B
10 / 13
Web Action Understanding
Gemma 4 26B (A4B)
4 / 4
Gemma 4 31B
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!