Atlas

Benchmarks

← All benchmarks

Arena Vision — Humor (No Style Control)

Multimodal

Raw Vision Arena rating for visually grounded humor prompts without style-control adjustment. Higher ratings indicate stronger preference in blind pairwise user votes.

Top models (higher is better)

ModelScore
Muse Spark1344
Claude Fable 51341
Gemini 3 Pro Preview1339
Gemini 3.1 Pro Preview1322
Gemini 3.5 Flash1321
Gemini 3.8 Flash1321
Claude Fable 5.11318
Opus 4.61314
Opus 4.71309
Qwen3.8-Max1304
Gemini 3.7 Flash1299
Claude Opus 51293
Muse Spark 1.31293
GPT-5.51293
Gemini 3 Flash Preview1291
Gemini 3.6 Flash1289
Muse Spark 1.11288
Gemini 3.5 Flash-Lite1287
GPT-5.41284
Kimi K2.51284
Opus 4.81282
Kimi K2.61280
Gemma 4 31B IT1280
GLM-5.3 Flash1279
Grok 4.51276
Gemma 4 26B A4B IT1272
Grok 4.61270
GLM 5V Turbo1268
Gemini 3.1 Flash-Lite Preview1266
GPT-5.11263
Gemini 2.5 Pro1262
Qwen3.5 397B A17B1262
Qwen3.7-Plus (source unspecified)1261
Sonnet 51261
DeepSeek V4.1 Flash1260
Gemini 2.5 Flash Preview (09-2025)1255
Sonnet 4.61248
GPT-5.6 Sol1246
Qwen3.5 122B A10B1243
Grok 4.1 Fast1239
Loading Atlas data…