AA-3D: leaderboard
Build interactive three.js scenes from 55 briefs in an offline Stirrup sandbox. Blind panel votes yield an unbounded Elo score. Preserve the model’s printed reasoning and fallback setting.
Metric: Elo. Source: artificialanalysis.ai. Saturation forecast: Around December 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra (Max) | 1165.24 | #5 (GPT-6 Astra) |
| 2 | GPT-6 Sol (Max) | 1122.68 | #14 (GPT-6 Sol) |
| 3 | GPT-6.1 Sol (Max) | 1114.47 | #8 (GPT-6.1 Sol) |
| 4 | Muse Spark 1.3 (Max) | 1035.28 | #25 (Muse Spark 1.3) |
| 5 | GPT-6 Luna (Max) | 1000 | #52 (GPT-6 Luna) |
| 6 | Claude Haiku 5.5 (Max) | 994.02 | #35 (Claude Haiku 5.5) |
| 7 | Grok 4.7 (xHigh) | 993.7 | #27 (Grok 4.7) |
| 8 | Kimi K3 (Max) | 981.87 | #30 (Kimi K3) |
| 9 | MiMo-V2.6-Pro | 977.24 | #37 |
| 10 | Gemini 3.8 Flash (High) | 964.84 | #21 (Gemini 3.8 Flash) |
| 11 | Qwen 3.8 Max (0902) | 948.27 | #28 |
| 12 | GLM-5.3 Flash | 930.66 | #66 |
| 13 | Step 5 Preview | 847.52 | #41 |
| 14 | GLM-5.3 (Max) | 781.69 | #40 (GLM-5.3) |
| 15 | MiniMax-M3 | 738.65 | #135 |
No result here: #3 Claude Opus 5.5, #7 Gemini 4 Argon, #9 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-3d · How It Works · Data refreshed daily, snapshot 2026-10-16.