Kimi k2.5 vs.Cosmos Reason2 8B
Passed:
Kimi k2.5 73.5% (50/68)
Cosmos Reason2 8B 63.4% (52/82)
Average request time:
Kimi k2.5 44.70s
Cosmos Reason2 8B 2.72s
Passed:
Kimi k2.5 73.5% (50/68)
Cosmos Reason2 8B 63.4% (52/82)
Average request time:
Kimi k2.5 44.70s
Cosmos Reason2 8B 2.72s
Summary
Annotation Understanding
Kimi k2.5
1 / 1
Cosmos Reason2 8B
2 / 2
CAPTCHA
Kimi k2.5
2 / 2
Cosmos Reason2 8B
2 / 2
Color Identification
Kimi k2.5
1 / 1
Cosmos Reason2 8B
1 / 1
Defect Detection
Kimi k2.5
8 / 14
Cosmos Reason2 8B
10 / 15
Document Understanding
Kimi k2.5
8 / 8
Cosmos Reason2 8B
5 / 10
Localization
Kimi k2.5
1 / 1
Cosmos Reason2 8B
1 / 1
OCR
Kimi k2.5
7 / 9
Cosmos Reason2 8B
8 / 9
Object Counting
Kimi k2.5
1 / 3
Cosmos Reason2 8B
3 / 11
Object Detection
Kimi k2.5
1 / 2
Cosmos Reason2 8B
0 / 2
Object Measurement
Kimi k2.5
0 / 1
Cosmos Reason2 8B
0 / 1
Object Understanding
Kimi k2.5
9 / 11
Cosmos Reason2 8B
8 / 11
Receipt Reading
Kimi k2.5
1 / 1
Cosmos Reason2 8B
1 / 1
Sign Understanding
Kimi k2.5
2 / 2
Cosmos Reason2 8B
2 / 2
Spatial Relations
Kimi k2.5
12 / 15
Cosmos Reason2 8B
10 / 17
Web Action Understanding
Kimi k2.5
3 / 4
Cosmos Reason2 8B
4 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!