Atlas

Benchmarks

← All benchmarks

Arena Text — Korean (No Style Control)

Chat & Writing

Raw Arena rating for Korean-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 Fable 51496
Claude Opus 51485
Gemini 3.5 Flash1477
GPT-5.51464
Gemini 3.1 Pro Preview1461
Muse Spark 1.11460
Muse Spark1460
Opus 4.71459
Opus 4.61458
Gemini 3 Pro Preview1450
GPT-5.41447
Gemini 3 Flash Preview1447
Gemini 2.5 Pro1435
GLM-5.21431
ERNIE 5.11431
MiMo-V2.5-Pro1429
Kimi K2.61428
Opus 4.51424
Qwen3.7-Plus1423
Opus 4.81423
GLM-5.11421
GPT-5.6 Sol1421
GLM-51420
Inkling1419
Sonnet 4.61414
DeepSeek-V4-Pro1410
Kimi K2.51410
Grok 4.11409
GLM-4.71407
Sonnet 51406
GPT-5.6 Terra1402
GLM 5V Turbo1400
Grok 4.51400
MiMo-V2-Pro1398
GPT-5.11396
Qwen3 Max Preview1395
Sonnet 4.51395
GPT-5.6 Luna1395
Gemini 3.1 Flash-Lite Preview1393
Qwen3 235B A22B Instruct 25071393
Loading Atlas data…