Atlas

Benchmarks

← All benchmarks

Arena Code WebDev — React

Code

Arena Code WebDev rating for websites and applications generated with React. Anonymous outputs are compared through blind pairwise user votes.

Top models (higher is better)

ModelScore
Claude Opus 51722
Kimi K31692
Claude Fable 51629
GLM-5.21597
DeepSeek-V4-Flash1590
Opus 4.81562
Opus 4.71558
Grok 4.51549
Sonnet 51547
Opus 4.61538
Hy31533
Gemini 3.6 Flash1528
Muse Spark 1.11527
Sonnet 4.61514
GLM-5.11510
Kimi K2.61505
MiniMax M31488
Gemini 3.5 Flash1486
Kimi K2.7 Code1472
MiMo-V2.5-Pro1463
Opus 4.51455
Qwen3.6 Plus (2026-04-02)1452
DeepSeek-V4-Pro1440
Gemini 3.1 Pro Preview1436
Gemini 3 Flash Preview1429
Kimi K2.51428
MiMo-V2.51426
MiMo-V2-Pro1425
GLM-4.71424
GLM-51424
Gemini 3.5 Flash-Lite1421
Inkling1404
GPT-5.3-Codex1399
Gemini 3 Pro Preview1396
MiniMax M2.71389
Qwen3.5 397B A17B1389
GPT-5.4 Mini1388
GPT-5.41386
GLM 5V Turbo1384
Sonnet 4.51384
Loading Atlas data…