Atlas

Benchmarks

← All benchmarks

Arena Text — Expert (No Style Control)

Chat & Writing

Raw Arena rating for expert 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 51564
Opus 4.61544
Claude Fable 51530
Opus 4.71521
GPT-5.51511
Gemini 3.5 Flash1511
GPT-5.41508
GPT-5.6 Terra1506
Opus 4.81503
GPT-5.6 Sol1502
MiMo-V2.5-Pro1502
Sonnet 4.61501
Kimi K31496
Gemini 3.6 Flash1495
Sonnet 51494
Qwen3.7 Max Preview1491
Kimi K2.61490
Opus 4.51487
ERNIE 5.11487
Gemini 3.1 Pro Preview1486
GLM-5.11481
Muse Spark 1.11481
Gemini 3 Pro Preview1480
Grok 4.51480
MiMo-V2-Pro1479
GPT-5.6 Luna1477
Sonnet 4.51472
Qwen3.7-Plus1471
Inkling1469
GPT-5.11468
GLM-5.21468
Kimi K2.51467
MiniMax M31466
DeepSeek-V4-Pro1466
Hy31466
Qwen3.5 397B A17B1466
Gemini 3 Flash Preview1463
MiMo-V2.51463
Qwen3 Max Preview1460
Qwen3.6 Plus (2026-04-02)1458
Loading Atlas data…