Atlas

Benchmarks

← All benchmarks

Arena Vision — Humor (Style Controlled)

Multimodal

Style-controlled Vision Arena rating for visually grounded humor prompts. Higher ratings indicate stronger preference in blind pairwise user votes after adjusting for response style.

Top models (higher is better)

ModelScore
Claude Fable 51324
Muse Spark1323
Gemini 3 Pro Preview1317
Gemini 3.1 Pro Preview1300
Gemini 3.8 Flash1298
Opus 4.61297
Gemini 3.5 Flash1293
Claude Fable 5.11291
Opus 4.71290
Qwen3.8-Max1290
GPT-5.51281
Gemini 3.5 Flash-Lite1278
Gemini 3.7 Flash1275
Muse Spark 1.31275
Muse Spark 1.11275
Grok 4.61271
Opus 4.81270
Gemini 3 Flash Preview1269
Gemini 3.6 Flash1268
Claude Opus 51265
GPT-5.11264
GPT-5.41264
Grok 4.51263
Kimi K2.51262
Kimi K2.61261
Gemma 4 31B IT1260
GPT-5.6 Sol1257
GLM-5.3 Flash1256
Gemini 3.1 Flash-Lite Preview1250
Sonnet 51249
Gemini 2.5 Pro1247
Gemma 4 26B A4B IT1245
DeepSeek V4.1 Flash1243
Qwen3.5 397B A17B1242
Sonnet 4.61241
Qwen3.7-Plus (source unspecified)1239
GPT-5.4 Mini1237
GPT-5.6 Terra1236
Grok 4.31229
Grok 4.1 Fast1227
Loading Atlas data…