Atlas

Benchmarks

← All benchmarks

Arena Text — Chinese (No Style Control)

Chat & Writing

Raw Arena rating for Chinese-language 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 51608
Opus 4.61554
Claude Fable 51545
GPT-5.51537
Opus 4.71536
Kimi K2.61531
Gemini 3.1 Pro Preview1531
Muse Spark 1.11530
Gemini 3.6 Flash1529
Gemini 3 Pro Preview1528
GPT-5.6 Sol1528
Qwen3.7 Max Preview1526
GLM-5.21522
Gemini 3.5 Flash1520
GPT-5.6 Terra1518
GPT-5.41517
GLM-51517
GLM-5.11517
Kimi K31516
Grok 4.51514
Gemini 3 Flash Preview1513
Qwen3.7-Plus1510
Opus 4.81509
Hy31508
Gemini 2.5 Pro1506
Muse Spark1505
ERNIE 5.11503
MiMo-V2.5-Pro1502
Qwen3.5 397B A17B1501
Kimi K2.51501
GLM 4.61500
Inkling1496
GLM 5V Turbo1493
GPT-5.11493
DeepSeek-V4-Pro1491
Sonnet 4.61489
Qwen3 Max Preview1487
GLM-4.71484
GPT-5.6 Luna1481
Gemini 3.5 Flash-Lite1479
Loading Atlas data…