Atlas

Benchmarks

← All benchmarks

Arena Code WebDev — HTML

Code

Arena Code WebDev rating for websites generated as HTML. Anonymous outputs are compared through blind pairwise user votes.

Top models (higher is better)

ModelScore
GPT-6.1 Sol1794
GPT-6 Astra1777
Claude Opus 5.51746
Claude Sonnet 5.51739
Muse Wonder Full1727
Kimi K3.1 Pro1702
DeepSeek V4.1 Flash 10171688
Muse Wonder1684
GPT-6 Sol1684
Claude Fable 5.11683
Claude Opus 51681
Muse Spark 1.31653
Qwen3.8 Max (0902)1652
Claude Fable 51650
Kimi K31642
Gemini 4 Argon1642
MiMo-V2.6-Flash1639
MiMo-V2.6-Pro1626
Qwen3.8-Max1625
Grok 4.71625
Hy4 Preview1620
GPT-5.6 Sol1616
Gemini 3.8 Flash1615
DeepSeek V4.1 Flash1613
Qwen3.8-Flash-Next1598
Grok 4.61597
GPT-6 Luna1596
GLM-5.31595
Gemini 3.7 Flash1592
GLM-5.3 Flash1577
Qwen3.8 27B1569
Grok 4.51567
GLM-5.21567
Claude Haiku 5.51566
DeepSeek V4 Pro 08131565
Opus 4.61559
DeepSeek-V4-Flash1558
Opus 4.71554
Gemini 3.5 Flash1554
Gemini 3.6 Flash1547
Loading Atlas data…