Atlas

Benchmarks

← All benchmarks

Arena Vision — Homework (Style Controlled)

Multimodal

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

Top models (higher is better)

ModelScore
GPT-6.1 Sol1358
Claude Fable 51352
GPT-6 Astra1342
Gemini 3.8 Flash1338
GPT-5.51335
Claude Fable 5.11334
Qwen3.8-Max1334
Opus 4.81332
Opus 4.71331
GPT-5.41330
Claude Sonnet 5.51326
Opus 4.61325
Grok 4.71324
Claude Opus 51323
GPT-5.6 Terra1321
Gemini 3.5 Flash1316
GPT-5.6 Sol1314
Gemini 3.1 Pro Preview1310
Gemini 3.7 Flash1309
Gemini 3 Pro Preview1309
Gemini 3 Flash Preview1308
DeepSeek V4.1 Flash1307
GLM-5.3 Flash1306
Sonnet 51305
Gemma 4 31B IT1304
Gemini 3.6 Flash1304
Kimi K2.61302
Sonnet 4.61302
Grok 4.51299
Muse Spark1299
Muse Spark 1.31297
Grok 4.61297
GPT-5.4 Mini1291
Gemini 3.5 Flash-Lite1290
Kimi K2.51290
GPT-5.6 Luna1290
Qwen3.7-Plus (source unspecified)1289
Muse Spark 1.11288
Qwen3.5 397B A17B1287
Gemma 4 26B A4B IT1285
Loading Atlas data…