Arena Vision · Diagrams

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

All results

98 rows

#	Model	Effort	Arena Vision · Diagrams
1	Fable 5	high	1353 ±10
2	Gemini 3.8 Flash	high	1333 ±18
3	Opus 4.7	high	1333 ±9
4	Opus 4.7	max	1331 ±9
5	Opus 4.6	max	1331 ±9
6	Opus 5.5	high	1330 ±23
7	Opus 5	high	1329 ±10
8	Opus 4.6	high	1324 ±9
9	Fable 5.1	max	1323 ±18
10	GPT-5.5	xhigh	1323 ±8
11	Qwen 3.8 Max	max	1323 ±12
12	Muse Spark 1.3	max	1322 ±17
13	Sonnet 5.5	xhigh	1321 ±24
14	GPT-5.4	high	1321 ±8
15	Gemini 3.7 Flash	high	1320 ±17
16	GPT-6.1 Sol	max	1319 ±25
17	Opus 4.8	high	1319 ±10
18	GPT-5.6 Sol	xhigh	1318 ±12
19	Muse Spark 1.2	xhigh	1316 ±26
20	Muse Spark 1.3	xhigh	1316 ±17
21	GPT-5.5	high	1315 ±8
22	GPT-5.4	xhigh	1314 ±9
23	Gemini 3.5 Flash	medium	1313 ±11
24	Muse Spark		1312 ±15
25	Gemini 3.6 Flash	high	1310 ±13
26	GPT-5.5 Instant		1310 ±14
27	GPT-6 Astra	max	1310 ±20
28	Gemini 3.1 Pro	high	1308 ±7
29	GPT-5.2	xhigh	1307 ±10
30	GLM 5.3 Flash	max	1306 ±10