Arena Vision · Creative writing

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

All results

91 rows

#	Model	Effort	Score
1	Gemini 3 Pro	high	1325 ±19
2	Fable 5	high	1319 ±18
3	GPT-6.1 Sol	max	1311 ±40
4	Opus 4.6	high	1309 ±17
5	Muse Spark		1303 ±33
6	Opus 4.7	high	1302 ±16
7	Gemini 3.1 Pro	high	1301 ±11
8	Opus 4.7	max	1298 ±17
9	Opus 4.6	max	1298 ±15
10	Gemini 3 Flash	high	1296 ±12
11	GPT-6 Astra	max	1294 ±31
12	Gemini 3.7 Flash	high	1292 ±28
13	Fable 5.1	max	1290 ±28
14	Opus 4.8	high	1289 ±17
15	GPT-5.6 Sol	xhigh	1289 ±20
16	Qwen 3.8 Max	max	1288 ±22
17	Opus 5.5	high	1287 ±37
18	Gemini 3 Flash	minimal	1283 ±13
19	GPT-5.5	high	1281 ±14
20	Grok 4.5	max	1281 ±19
21	Opus 4.8	max	1280 ±17
22	Gemini 3.5 Flash	medium	1277 ±19
23	Gemini 3.8 Flash	high	1276 ±29
24	Opus 5	high	1275 ±16
25	Gemini 3.5 Flash	high	1275 ±19
26	Muse Spark 1.3	max	1272 ±28
27	GPT-5.5 Instant		1271 ±32
28	GPT-5.5	xhigh	1269 ±14
29	Muse Spark 1.3	xhigh	1268 ±28
30	Sonnet 4.6	max	1266 ±15