LMArena Vision

crowdrun by LMArenaunit: eloCC-BY-4.0observed Jul 21, 202646 modelssource ↗methodology ↗

Caveat: Image-understanding battles.

#	Model	Score
1	Claude Fable 5Anthropic	1334 ±9· 7.4K votes
2	Claude Opus 4.7thinkingAnthropic	1320 ±7· 19.4K votes
3	Claude Opus 4.6thinkingAnthropic	1315 ±7· 19.2K votes
4	Gemini 3.5 Flashhigh-effortGoogle	1301 ±12· 3K votes
5	GPT-5.4high-effortOpenAI	1299 ±7· 19.4K votes
6	GPT-5.5OpenAI	1296 ±7· 17.4K votes
7	Gemini 3.1 Pro PreviewGoogle	1295 ±6· 35.8K votes
8	Claude Opus 4.8thinkingAnthropic	1294 ±8· 10K votes
9	Kimi K2.6Moonshot AI	1283 ±7· 15.6K votes
10	Claude Sonnet 5high-effortAnthropic	1283 ±10· 4.9K votes
11	Claude Sonnet 4.6Anthropic	1281 ±7· 24K votes
12	Qwen3.7 PlusAlibaba	1276 ±9· 6.2K votes
13	GPT-5.2 ChatOpenAI	1267 ±7· 15.6K votes
14	Qwen3.5 397B-A17BAlibaba	1267 ±6· 22.9K votes
15	Kimi K2.5thinkingMoonshot AI	1266 ±6· 25.2K votes
16	GLM-5V-TurboZhipu AI	1266 ±7· 26.8K votes
17	Gemini 2.5 ProGoogle	1261 ±5· 89.4K votes
18	MiniMax-M3MiniMax (minimax.io)	1257 ±8· 9.8K votes
19	GPT-5.1high-effortOpenAI	1248 ±8· 9.2K votes
20	GPT-5.4 minihigh-effortOpenAI	1247 ±7· 22.3K votes
21	GPT-5.2high-effortOpenAI	1246 ±6· 18.1K votes
22	Qwen3.5 122B-A10BAlibaba	1246 ±7· 12.6K votes
23	Qwen3.5 27BAlibaba	1241 ±7· 20.6K votes
24	Gemini 3.1 Flash Lite PreviewGoogle	1240 ±6· 26.3K votes
25	Gemini 2.5 FlashGoogle	1236 ±5· 59.8K votes
26	Grok 4.3xAI	1233 ±7· 16.8K votes
27	o3OpenAI	1215 ±7· 46.6K votes
28	Qwen-VL MaxAlibaba	1214 ±12· 3.2K votes
29	GPT-5high-effortOpenAI	1211 ±7· 35K votes
30	GPT-4.1OpenAI	1210 ±7· 42K votes
31	Qwen3-VL 235B-A22BthinkingAlibaba	1207 ±12· 2.4K votes
32	GPT-5 Minihigh-effortOpenAI	1203 ±8· 29.6K votes
33	GPT-5.4 nanohigh-effortOpenAI	1202 ±7· 22.3K votes
34	o4-miniOpenAI	1195 ±7· 42.3K votes
35	GPT-4.1 miniOpenAI	1182 ±7· 41.3K votes
36	o1OpenAI	1168 ±11· 3.7K votes
37	GLM-4.6VZhipu AI	1165 ±14· 2.3K votes
38	GPT-5 Nanohigh-effortOpenAI	1162 ±11· 4K votes
39	GLM-4.5VZhipu AI	1156 ±12· 3.3K votes
40	GPT-4oOpenAI	1137 ±9· 23.3K votes
41	Qwen2.5-VL 72B InstructAlibaba	1107 ±11· 3.8K votes
42	GPT-4 TurboOpenAI	1090 ±11· 13.4K votes
43	GPT-4o miniOpenAI	1066 ±8· 17.3K votes
44	GPT-4.1 nanoOpenAI	1063 ±18· 1.2K votes
45	Pixtral 12BMistral	1009 ±9· 7.5K votes
46	Aya Vision 32BCohere	996 ±22· 847 votes

Dot = elo on a 955–1362 scale (position, not length); whisker = rating confidence interval. Scores link to model pages; every number keeps its source on the model page.