Benchmarks · composite

Arena Text · Spanish

Crowdsourced pairwise human preference on open-ended text prompts (style control on); Bradley-Terry rating. Only the Spanish category.

All results

196 rows

#	Model	Effort	Arena Text · Spanish
1	Gemini 4 Argon	high	1506 ±22
2	Opus 4.7	high	1498 ±14
3	Opus 4.6	max	1496 ±12
4	Fable 5	high	1495 ±16
5	Muse Spark 1.2	xhigh	1495 ±27
6	Fable 5.1	max	1495 ±23
7	Opus 4.6	high	1493 ±12
8	Opus 5.5	high	1493 ±27
9	Muse Spark 1.3	max	1491 ±21
10	Opus 5	max	1490 ±19
11	GPT-5.5 Instant		1485 ±22
12	Kimi K3	max	1484 ±17
13	GLM 5.2	max	1484 ±15
14	Gemini 3.1 Pro	high	1482 ±10
15	Opus 4.8	high	1482 ±15
16	Qwen 3.8 Max	max	1480 ±18
17	Gemini 3.8 Flash	high	1480 ±14
18	Gemini 3.7 Flash	high	1479 ±19
19	Muse Wonder	max	1479 ±28
20	Gemini 3 Flash	high	1478 ±19
21	Muse Spark		1478 ±27
22	Opus 5	high	1477 ±14
23	Opus 4.5	high	1477 ±18
24	Gemini 3.5 Flash	high	1474 ±17
25	Opus 4.7	max	1474 ±14
26	Muse Spark 1.1	max	1473 ±16
27	Gemini 3.6 Flash	high	1473 ±20
28	GPT-5.2	xhigh	1473 ±18
29	GLM 5.3 Flash	max	1472 ±15
30	GPT-6.1 Sol	max	1472 ±28