Atlas

Benchmarks

← All benchmarks

Arena Text — Instruction Following (No Style Control)

Chat & Writing

Raw Arena rating for instruction-following Text Arena prompts without style-control adjustment. Higher ratings indicate stronger preference in blind pairwise user votes.

Top models (higher is better)

ModelScore
Claude Opus 5.51539
Gemini 4 Argon1528
Opus 4.61521
Claude Fable 5.11519
Claude Opus 51518
Claude Fable 51502
Opus 4.71498
Muse Wonder1493
Gemini 3.8 Flash1492
MiMo-V2.6-Pro1491
Claude Sonnet 5.51491
Gemini 3.7 Flash1486
Kimi K31485
GPT-5.6 Sol1483
Qwen3.8-Max1481
GLM-5.31480
GLM-5.3 Flash1479
Muse Spark 1.31479
GPT-5.51478
Opus 4.81476
Sonnet 4.61476
MiMo-V2.5-Pro1475
DeepSeek V4.1 Flash1474
Opus 4.51472
GPT-5.41469
Gemini 3.6 Flash1467
GPT-6.1 Sol1467
Gemini 3.5 Flash1467
Gemini 3.1 Pro Preview1466
GLM-5.21466
Muse Spark 1.21463
Qwen3.7 Max Preview1460
ERNIE 5.11460
Sonnet 4.51459
Gemini 3 Pro Preview1458
Muse Spark 1.11456
GPT-5.6 Terra1455
GPT-6 Astra1455
GLM-5.11452
Sonnet 51452
Loading Atlas data…