Arena Vision · Humor

Human preference on prompts with image inputs (multimodal understanding); Bradley-Terry rating. Only the Humor category.

All results

83 rows

#	Model	Effort	Arena Vision · Humor
1	Fable 5	high	1324 ±25
2	Muse Spark		1323 ±40
3	Gemini 3 Pro	high	1317 ±25
4	Gemini 3.1 Pro	high	1300 ±14
5	Gemini 3.8 Flash	high	1298 ±46
6	Opus 4.6	high	1297 ±21
7	Opus 4.6	max	1295 ±19
8	Gemini 3.5 Flash	high	1293 ±25
9	Fable 5.1	max	1291 ±41
10	Opus 4.7	high	1290 ±21
11	Qwen 3.8 Max	max	1290 ±29
12	GPT-5.5 Instant		1287 ±40
13	Opus 4.7	max	1286 ±20
14	Gemini 3.5 Flash	medium	1286 ±26
15	GPT-5.5	high	1281 ±19
16	Gemini 3.5 Flash-Lite	xhigh	1278 ±30
17	Gemini 3.7 Flash	high	1275 ±41
18	Muse Spark 1.3	xhigh	1275 ±42
19	Muse Spark 1.1	max	1275 ±26
20	Grok 4.6	high	1271 ±33
21	Opus 4.8	high	1270 ±22
22	Gemini 3 Flash	high	1269 ±16
23	Gemini 3.6 Flash	high	1268 ±30
24	Opus 4.8	max	1267 ±22
25	Muse Spark 1.3	max	1267 ±39
26	GPT-5.5	xhigh	1266 ±18
27	Opus 5	high	1265 ±21
28	GPT-5.1	high	1264 ±32
29	Grok 4.20	xhigh	1264 ±20
30	GPT-5.4	high	1264 ±19