One fixed creative coding challenge
One prompt.
Many capybaras.
Each model gets the same prompt: animate a capybara on a hoverboard. Browse the outputs, then inspect the evidence.
Collected attempts 10
One task, not an overall model ranking. Original evidence is preserved.
Filter and sort
10 attempts
Inspect gpt-oss-120b-medium
- Model time
- 10.721 s
- Total tokens
- 13,653
Opt-in live preview
- Time
- Unknown
- Total tokens
- 3,163,198
Opt-in live preview
- Model time
- 980.485 s
- Total tokens
- 5,464,687
Opt-in live preview
- Model time
- 575.1 s
- Total tokens
- 2,156,080
Opt-in live preview
- Model time
- 317.707 s
- Total tokens
- 344,028
Opt-in live preview
- Model time
- 81.426 s
- Total tokens
- 12,457
Opt-in live preview
- Model time
- 52.654 s
- Total tokens
- 7,995
Opt-in live preview
- Model time
- 145.973 s
- Total tokens
- 15,897
Opt-in live preview
- Model time
- 99.841 s
- Total tokens
- 10,597
Opt-in live preview
- Model time
- 17.222 s
- Total tokens
- 3,726
Opt-in live preview
No matching attempts
Try another search or reset your filters.
Inspired by Simon Willison's pelican-riding-a-bicycle experiments.
