Benchmark Prompts / P01
Treasure chest: 11 AI models, one prompt
Every Season 3 competitor received this prompt verbatim, at default settings, with one generation and no manual cleanup. Nothing below is cherry-picked — the failures ship too, because a benchmark that hides them isn't one.
11Outputs580 ballots · 13 both-bad
A wooden treasure chest with iron fittings and a curved lid, slightly weathered, closed.
Every output for this prompt
Click a tile to load the model in 3D, then switch to Wire or X-ray to read the topology. Vote on these blind →
Claude Opus 5.5Treasure chest Grok 4.7Treasure chest GPT-6 SolTreasure chest GPT-6 LunaTreasure chest GPT-6.1 SolTreasure chest 3D-AgentTreasure chest GPT-6 AstraTreasure chest Claude Fable 5.1Treasure chest Claude Opus 5Treasure chest GPT-5.6 SolTreasure chest Claude Fable 5Treasure chest
Season 3 blind votes for P01
580 ballots · 13 “both bad”
| Competitor | Record | Wins | Losses |
|---|---|---|---|
| Claude Opus 5.5 | 99 | 14 | |
| 3D-Agent | 81 | 31 | |
| Claude Fable 5.1 | 73 | 37 | |
| GPT-5.6 Sol | 64 | 37 | |
| Claude Opus 5 | 55 | 58 | |
| GPT-6 Astra | 46 | 57 | |
| GPT-6.1 Sol | 45 | 16 | |
| GPT-6 Sol | 42 | 31 | |
| GPT-6 Luna | 23 | 71 | |
| Claude Fable 5 | 18 | 105 | |
| Grok 4.7 | 5 | 94 |