llm.ing

Design Arena — spreadsheet

mirrorrun by Design Arenaunit: elo30 modelssource ↗

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

websitecodecategoriesgamedev3duicomponentdatavizsvgasciiartmobileappswebapps+19 more
#ModelScoreObserved
1Claude Fable 5.2
1430
Nov 19, 2026
2Grok 4.7
1420
Nov 19, 2026
3Claude Fable 5.1
1344
Nov 19, 2026
4Kimi K3.1 Pro
1318
Nov 19, 2026
5GLM-5.1
1315
Oct 19, 2026
6Claude Opus 5.5
1292
Nov 19, 2026
7Grok 4.6
1280
Nov 19, 2026
8Kimi K2.6
1276
Oct 19, 2026
9Claude Sonnet 5.5
1251
Nov 19, 2026
10Claude Opus 5
1242
Nov 19, 2026
11Gemini 4 Carbon
1224
Nov 19, 2026
12GPT-6.1 Sol
1209
Nov 19, 2026
13MiniMax-M3
1168
Nov 19, 2026
14Gemini 4 Argon
1145
Nov 19, 2026
15Muse Spark 1.3
1127
Nov 19, 2026
16Grok 4.5
1059
Nov 19, 2026
17Kimi K3
1054
Nov 19, 2026
18Gemini 3.6 Flash
1049
Nov 19, 2026
19Claude Opus 4.8
1036
Nov 19, 2026
20Claude Sonnet 5
1033
Nov 19, 2026
21Gemini 3.8 Flash
1026
Nov 19, 2026
22Gemini 3.7 Flash
1018
Nov 19, 2026
23GLM-5.2
1013
Nov 19, 2026
24Muse Spark 1.2
1005
Nov 19, 2026
25Muse Spark 1.1
1001
Nov 19, 2026
26Qwen3.7 Max
993
Nov 19, 2026
27Claude Opus 4.6
989
Nov 19, 2026
28Gemini 3.5 Flash
982
Nov 19, 2026
29Grok 4.3
923
Nov 19, 2026
30Gemini 3.1 Pro Preview
918
Nov 19, 2026

Dot = elo on a 892–1456 scale (position, not length). Scores link to model pages; every number keeps its source on the model page.