Design Arena (Game Dev): leaderboard
Crowdsourced arena for AI game development. Models generate browser games and are ranked by Elo from human votes.
Metric: Elo. Source: www.designarena.ai. Status: saturation imminent. 161 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5.1 | 1430 |
| 2 | Kimi K3 | 1419 |
| 3 | GPT-5.6 Sol (xHigh) | 1397 |
| 4 | Claude Fable 5 | 1373 |
| 5 | Claude Opus 5 | 1373 |
| 6 | GPT-5.6 Sol | 1368 |
| 7 | Gemini 3.7 Flash | 1345 |
| 8 | Gemini 3.8 Flash | 1344 |
| 9 | Muse Spark 1.2 | 1340 |
| 10 | GPT-5.5 | 1329 |
| 11 | Claude Sonnet 5 | 1321 |
| 12 | Claude Opus 4.7 (Thinking) | 1320 |
| 13 | GLM-5.2 | 1311 |
| 14 | Claude Opus 4.6 (Thinking) | 1309 |
| 15 | Claude Opus 4.6 | 1302 |
Interactive version: theaggregate.ai/benchmark?slug=design-arena-game-dev · How It Works · Data refreshed daily, snapshot 2026-09-05.