Atlas

Benchmarks

← All benchmarks

Arena Text — Japanese (No Style Control)

Chat & Writing

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

Top models (higher is better)

ModelScore
Gemini 3 Pro Preview1513
Claude Fable 51507
GPT-5.51504
Gemini 3.1 Pro Preview1498
Gemini 3 Flash Preview1495
GPT-5.41482
Opus 4.61479
Opus 4.71474
Gemini 2.5 Pro1465
GLM-5.21464
Gemini 3.5 Flash1462
GPT-5.11455
DeepSeek-V4-Pro1455
Kimi K2.61446
Sonnet 51441
Qwen3.5 397B A17B1431
Opus 4.81426
Sonnet 4.61424
Kimi K2.51423
ERNIE 5.11423
GLM-5.11421
GPT-5.21417
GLM-4.51417
MiMo-V2.5-Pro1414
Gemini 3.1 Flash-Lite Preview1413
Opus 4.51412
Gemini 2.5 Flash Preview (09-2025)1407
O3 (2025-04-16)1406
GLM-51405
Qwen3 Next 80B A3B Instruct1405
Qwen3.7-Plus1402
Grok 4.11402
Qwen3 Max Preview1399
Gemini 2.5 Flash1398
GPT-51397
MiniMax M31395
GLM 4.61393
Sonnet 4.51389
Qwen3 235B A22B Instruct 25071389
Grok 41388
Loading Atlas data…