Atlas

Benchmarks

← All benchmarks

Arena Vision — Diagram (Style Controlled)

Multimodal

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

Top models (higher is better)

ModelScore
Claude Fable 51353
Gemini 3.8 Flash1333
Opus 4.71333
Opus 4.61331
Claude Opus 51329
Claude Fable 5.11323
GPT-5.51323
Qwen3.8-Max1323
Muse Spark 1.31322
Claude Sonnet 5.51321
GPT-5.41321
Gemini 3.7 Flash1320
GPT-6.1 Sol1319
Opus 4.81319
GPT-5.6 Sol1318
Muse Spark 1.21316
Gemini 3.5 Flash1313
Muse Spark1312
Gemini 3.6 Flash1310
GPT-6 Astra1310
Gemini 3.1 Pro Preview1308
GLM-5.3 Flash1306
Grok 4.51306
Sonnet 4.61306
Muse Spark 1.11305
Gemini 3 Pro Preview1301
Step 5 Preview1301
GPT-5.6 Terra1299
Grok 4.61294
DeepSeek V4.1 Flash1293
Sonnet 51293
Kimi K2.61293
Gemini 3 Flash Preview1293
Gemini 3.5 Flash-Lite1292
Qwen3.7-Plus (source unspecified)1290
GPT-5.4 Mini1283
Gemma 4 31B IT1283
Grok 4.71280
GPT-5.6 Luna1280
Kimi K2.51280
Loading Atlas data…