llm.ing

Design Arena — agenticgamedev

mirrorrun by Design Arenaunit: elo25 modelssource ↗

Caveat: Head-to-head human votes on generated designs/code, mirrored via OpenRouter.

websitecodecategoriesdatavizuicomponentgamedev3dsvgasciiartfullstackmobileapps+15 more
#ModelScoreObserved
1Claude Fable 5
1285
Sep 12, 2026
2Claude Opus 5
1261
Sep 14, 2026
3Kimi K3
1252
Sep 12, 2026
4Claude Opus 4.7
1247
Sep 5, 2026
5Claude Opus 4.8
1238
Sep 12, 2026
6Grok 4.5
1235
Sep 4, 2026
6Gemini 3.7 Flash
1235
Sep 6, 2026
8Claude Sonnet 5
1227
Sep 20, 2026
9Claude Opus 4.6
1215
Sep 12, 2026
10Grok 4.6
1210
Sep 8, 2026
11GLM-5.2
1205
Sep 12, 2026
12Muse Spark 1.1
1183
Sep 12, 2026
13Gemini 3.5 Flash
1180
Aug 31, 2026
13Gemini 3.6 Flash
1180
Sep 12, 2026
15GPT-5.5
1179
Sep 12, 2026
15Muse Spark 1.2
1179
Sep 8, 2026
17Claude Sonnet 4.6
1177
Sep 12, 2026
18GLM-5.1
1165
Sep 2, 2026
19Qwen3.7 Max
1161
Sep 6, 2026
20MiniMax-M3
1158
Sep 20, 2026
21Kimi K2.6
1141
Sep 12, 2026
22Kimi K2.7 Code
1130
Sep 6, 2026
23Gemini 3.1 Pro Preview
1114
Sep 8, 2026
24GLM-5V-Turbo
1102
Sep 11, 2026
25Grok 4.3
1007
Sep 12, 2026

Dot = elo on a 993–1299 scale (position, not length). Scores link to model pages; every number keeps its source on the model page.