Atlas

Benchmarks

← All benchmarks

Arena Vision — Entity Recognition (Style Controlled)

Multimodal

Style-controlled Vision Arena rating for visual entity-recognition prompts. Higher ratings indicate stronger preference in blind pairwise user votes after adjusting for response style.

Top models (higher is better)

ModelScore
Gemini 3 Pro Preview1292
Gemini 3 Flash Preview1290
Gemini 3.1 Pro Preview1284
Gemini 3.1 Flash-Lite Preview1271
GPT-51261
Gemini 2.5 Pro1251
Opus 4.71242
Opus 4.61237
Grok 41236
Kimi K2.51232
o31231
GPT-5.11228
Gemini 2.5 Flash1221
GPT-5.51219
O4 Mini1219
Grok 4.1 Fast1217
GPT-5.41215
Qwen3.5 397B A17B1211
GPT-5 Mini1211
Grok 4.31210
GPT-5.21201
GPT-5.4 Mini1198
GPT-4.1 Mini1197
Gemma 4 31B IT1196
GPT-4.11184
Gemini 2.5 Flash-Lite1183
GLM 5V Turbo1175
Gemma 3 27B IT1170
Sonnet 4.61167
MiMo-V2.51163
MiniMax M31161
Gemma 4 26B A4B IT1158
Mistral Small 3.2 24B Instruct 25061134
Mistral Small 3.1 24B Instruct 25031131
Mistral Medium 3.11128
Mistral Medium 31125
GPT-5.4 Nano1116
Loading Atlas data…