Atlas

Benchmarks

← All benchmarks

Arena Code WebDev — Gaming

Code

Arena Code WebDev rating for interactive browser games. Anonymous generated games are compared through blind pairwise user votes.

Top models (higher is better)

ModelScore
Claude Opus 5.51929
Claude Sonnet 5.51884
GPT-6 Astra1861
Claude Fable 5.11859
GPT-6.1 Sol1833
Muse Wonder Full1797
Claude Opus 51764
Gemini 4 Argon1763
GPT-6 Sol1751
Qwen3.8 Max (0902)1745
Qwen3.8-Max1737
Muse Spark 1.31731
Kimi K3.1 Pro1724
Kimi K31706
MiMo-V2.6-Flash1704
DeepSeek V4.1 Flash 10171702
Hy4 Preview1698
Claude Fable 51695
Grok 4.71694
Qwen3.8-Flash-Next1693
Muse Wonder1689
MiMo-V2.6-Pro1681
Claude Haiku 5.51679
Grok 4.61674
GPT-5.6 Sol1672
GLM-5.31671
DeepSeek V4.1 Flash1669
MiniMax M3.1 Flash1654
GLM-5.3 Flash1650
GLM-5.21649
Gemini 3.7 Flash1638
Gemini 3.8 Flash1635
Qwen3.8 27B1634
GPT-6 Luna1614
Opus 4.81607
DeepSeek V4 Pro 08131606
DeepSeek-V4-Flash1606
Opus 4.71588
Grok 4.51587
Opus 4.61584
Loading Atlas data…