Atlas

Benchmarks

← All benchmarks

Arena Vision — Entity Recognition (No Style Control)

Multimodal

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

Top models (higher is better)

ModelScore
Gemini 3 Pro Preview1348
Gemini 3 Flash Preview1336
Gemini 3.1 Pro Preview1330
Gemini 3.1 Flash-Lite Preview1308
Grok 41288
Kimi K2.51285
Gemini 2.5 Pro1283
Opus 4.71278
Opus 4.61274
Qwen3.5 397B A17B1265
GPT-51263
Gemini 2.5 Flash1252
Grok 4.1 Fast1251
GPT-5.41251
GLM 5V Turbo1241
o31234
GPT-5.21232
GPT-5.11232
GPT-5.51231
GPT-5 Mini1226
Gemma 4 31B IT1226
GPT-5.4 Mini1218
O4 Mini1218
Grok 4.31217
GPT-4.11208
MiniMax M31197
Gemma 3 27B IT1196
Sonnet 4.61195
Gemma 4 26B A4B IT1191
MiMo-V2.51189
GPT-4.1 Mini1187
Gemini 2.5 Flash-Lite1186
Mistral Small 3.1 24B Instruct 25031155
Mistral Small 3.2 24B Instruct 25061151
Mistral Medium 3.11147
GPT-5.4 Nano1141
Mistral Medium 31141
Loading Atlas data…