Atlas

Benchmarks

← All benchmarks

Arena Text — Multi-turn (No Style Control)

Chat & Writing

Raw Arena rating for multi-turn Text Arena conversations without style-control adjustment. Higher ratings indicate stronger preference in blind pairwise user votes.

Top models (higher is better)

ModelScore
Gemini 4 Argon1547
Opus 4.61512
Muse Wonder1507
Opus 4.71505
Claude Fable 51504
Claude Opus 5.51504
Gemini 3.8 Flash1503
Claude Opus 51500
Muse Spark 1.21500
Claude Fable 5.11495
Qwen3.8-Max1490
Kimi K31489
Muse Spark 1.31489
Gemini 3.1 Pro Preview1487
Gemini 3.7 Flash1487
Gemini 3 Pro Preview1484
Muse Spark 1.11484
GPT-5.41482
Gemini 3.6 Flash1481
Gemini 3.5 Flash1481
Qwen3.7 Max Preview1480
Muse Spark1477
GPT-5.51476
Opus 4.81476
MiMo-V2.5-Pro1475
GLM-5.11472
Gemini 3 Flash Preview1471
GLM-5.3 Flash1471
ERNIE 5.11470
GLM-5.31470
GLM-5.21469
DeepSeek-V4-Pro1468
Opus 4.51466
Claude Sonnet 5.51465
MiMo-V2.6-Pro1465
Sonnet 4.51465
Sonnet 4.61464
GPT-5.6 Sol1461
DeepSeek V4.1 Flash1460
Qwen3.7-Plus (source unspecified)1459
Loading Atlas data…