Atlas

Benchmarks

← All benchmarks

Arena Vision — OCR (No Style Control)

Multimodal

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

Top models (higher is better)

ModelScore
Claude Fable 51337
Gemini 3.8 Flash1333
Claude Opus 51329
Claude Fable 5.11327
Opus 4.71327
Opus 4.61326
Qwen3.8-Max1322
Gemini 3.7 Flash1322
Muse Spark 1.31319
Gemini 3.5 Flash1317
Muse Spark 1.21313
GPT-5.41312
Gemini 3.6 Flash1312
Gemini 3 Pro Preview1310
Muse Spark1307
GPT-5.51306
Gemini 3.1 Pro Preview1304
Opus 4.81303
Muse Spark 1.11301
GLM-5.3 Flash1299
Claude Sonnet 5.51297
Sonnet 4.61296
Grok 4.51293
GPT-6 Astra1292
Kimi K2.61291
Gemini 3 Flash Preview1291
GPT-5.6 Sol1290
GPT-6.1 Sol1289
Qwen3.7-Plus (source unspecified)1288
Sonnet 51288
DeepSeek V4.1 Flash1282
Step 5 Preview1282
Gemma 4 31B IT1282
Qwen3.8 27B1281
Gemini 3.5 Flash-Lite1280
GPT-5.6 Terra1278
Kimi K2.51276
Grok 4.61272
Qwen3.5 397B A17B1271
Gemini 2.5 Pro1270
Loading Atlas data…