Atlas

Benchmarks

← All benchmarks

Arena Vision — Creative Writing (No Style Control)

Multimodal

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

Top models (higher is better)

ModelScore
Claude Fable 51352
Gemini 3 Pro Preview1349
Opus 4.61338
Muse Spark1332
Opus 4.71326
Gemini 3.1 Pro Preview1320
Opus 4.81316
Gemini 3.5 Flash1312
GPT-5.51298
Gemini 3 Flash Preview1298
Claude Opus 51284
Gemma 4 31B1280
Qwen3.5 397B A17B1278
GPT-5.41277
Sonnet 4.61273
Kimi K2.61271
Gemini 2.5 Pro1270
Gemini 2.5 Flash Preview (09-2025)1264
Qwen3.7-Plus1264
Kimi K2.51263
Gemma 4 26B A4B1261
Sonnet 51260
Muse Spark 1.11258
GPT-5.11255
GLM 5V Turbo1253
Gemini 3.1 Flash-Lite Preview1249
Grok 4.31249
Gemini 2.5 Flash1246
Grok 4.1 Fast1245
Grok 41239
MiniMax M31230
Qwen3.5 122B A10B1229
MiMo-V2.51227
GPT-5.21225
GPT-4.1 (2025-04-14)1222
GPT-5.4 Mini1220
O3 (2025-04-16)1214
Mistral Medium 3.11212
GPT-51202
Gemini 2.5 Flash-Lite Preview (09-2025)1201
Loading Atlas data…