Benchmarks · creative
threejseval
Coding models build a Three.js scene from the same prompt; humans vote blind on pairs of finished scenes. Value is Bradley-Terry Elo (base 1500), the mean across prompts.
30 models · 38 results · updated 2026-10-28 · threejseval.com ↗
Leaderboard
Opus 5.5 high
More from this benchmark
3 published measuresScore vs any published measure3 measures: sub-scores, splits, counts
All results
38 rows| # | Model | Effort | threejseval |
|---|---|---|---|
| 1 | Opus 5.5 | high | 2065 |
| 2 | GPT-6.1 Sol | high | 1976 |
| 3 | GPT-6 Astra | medium | 1883 |
| 4 | Opus 5.5 | medium | 1882 |
| 5 | Fable 5.1 | high | 1860 |
| 6 | Sonnet 5.5 | high | 1825 |
| 7 | DeepSeek V4.1 Flash 1017 | max | 1735 |
| 8 | Gemini 4 Argon | high | 1718 |
| 9 | Opus 5 | medium | 1709 |
| 10 | DeepSeek V4.1 Flash | max | 1633 |
| 11 | GPT-6 Sol | high | 1627 |
| 12 | Grok 4.7 | xhigh | 1619 |
| 13 | Grok 4.7 | high | 1597 |
| 14 | MiniMax M3.1 | max | 1578 |
| 15 | Grok 4.7 | medium | 1573 |
| 16 | Gemini 3.8 Flash | high | 1566 |
| 17 | Kimi K3.1 | high | 1529 |
| 18 | Fable 5 | medium | 1516 |
| 19 | Kimi K3 | high | 1511 |
| 20 | Gemini 3.7 Flash | high | 1467 |
| 21 | Kimi K3 | low | 1458 |
| 22 | Grok 4.6 | high | 1454 |
| 23 | Muse Spark 1.3 | max | 1436 |
| 24 | GPT-5.6 Terra | max | 1432 |
| 25 | Grok 4.7 | low | 1423 |
| 26 | Haiku 5.5 | high | 1413 |
| 27 | GPT-5.6 Sol | medium | 1369 |
| 28 | GPT-6 Luna | high | 1365 |
| 29 | Gemini 3.6 Flash | high | 1341 |
| 30 | Grok 4.6 | medium | 1285 |