Atlas

Benchmarks

← All benchmarks

Arena Vision — Chinese (No Style Control)

Multimodal

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

Top models (higher is better)

ModelScore
Gemini 3.5 Flash1400
Claude Fable 51387
Opus 4.61379
Gemini 3 Pro Preview1376
Muse Spark 1.11372
Gemini 3.1 Pro Preview1369
Opus 4.71365
Sonnet 51358
GPT-5.51356
GPT-5.41355
Muse Spark1354
Kimi K2.61349
Opus 4.81345
Qwen3.7-Plus1337
Gemini 3 Flash Preview1337
GLM 5V Turbo1335
Gemma 4 31B1332
MiMo-V2.51327
MiniMax M31321
Qwen3.5 397B A17B1321
Gemma 4 26B A4B1315
Kimi K2.51314
Sonnet 4.61306
Qwen3.5 122B A10B1301
GPT-5.4 Mini1296
GPT-5.21294
Gemini 2.5 Flash Preview (09-2025)1291
Gemini 2.5 Pro1290
Gemini 2.5 Flash1285
Gemini 3.1 Flash-Lite Preview1284
MiMo-V2-Omni1280
GPT-5.11279
Grok 4.31263
Grok 41253
GPT-51239
GPT-5.4 Nano1233
Grok 4.1 Fast1232
Gemini 2.5 Flash-Lite Preview (09-2025)1231
O3 (2025-04-16)1231
GPT-4.1 (2025-04-14)1229
Loading Atlas data…