Atlas

Benchmarks

← All benchmarks

Arena Vision — English (No Style Control)

Multimodal

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

Top models (higher is better)

ModelScore
Opus 4.61324
Claude Fable 51323
Opus 4.71322
Gemini 3.8 Flash1318
Claude Opus 51315
Claude Fable 5.11313
Qwen3.8-Max1311
Gemini 3.7 Flash1309
Gemini 3.5 Flash1305
Muse Spark1304
GPT-5.41301
Sonnet 4.61298
Muse Spark 1.31298
GLM-5.3 Flash1295
Gemini 3.6 Flash1295
Opus 4.81293
GPT-5.51292
Muse Spark 1.11292
Gemini 3 Pro Preview1292
Muse Spark 1.21291
Step 5 Preview1290
Claude Sonnet 5.51289
DeepSeek V4.1 Flash1288
Gemini 3.1 Pro Preview1287
Kimi K2.61287
Grok 4.51285
Qwen3.7-Plus (source unspecified)1280
GPT-5.6 Sol1280
GPT-6 Astra1280
Gemini 3 Flash Preview1276
Sonnet 51275
GLM 5V Turbo1274
Kimi K2.51274
GPT-6.1 Sol1274
GPT-5.6 Terra1272
Gemma 4 31B IT1271
Grok 4.61270
MiMo-V2.6-Flash1269
Gemini 3.5 Flash-Lite1269
Qwen3.5 397B A17B1268
Loading Atlas data…