111 playgrounds · mean 98.7/100 · 9 validated · 84 reproduce · 0 honesty failures · citations not checked · back to playgrounds
score is one 0 to 100 number: how well a playground conforms to the template, weighting build, registration, structure, scientific infrastructure and house rules into a single figure. hover a score for the per-category breakdown. honesty: cal ✓✓ validated against an external target, cal ✓ computes and reproduces its own numbers, cal ~ a declared showcase of a poor model; a failure caps the score.
model leaderboard
| # | model | n | mean | |
|---|---|---|---|---|
| 1 | Claude Opus 4.8 | 98 | 100 | |
| 2 | Claude Opus 5 | 2 | 100 | |
| 3 | Claude Fable 5 | 5 | 100 | |
| 4 | Codex Gpt-5 | 1 | 100 | |
| 5 | Claude Opus 5.5 | 2 | 100 | |
| (unattributed) | 3 | no versions.ts declared | ||