Atlas

Benchmarks

← All benchmarks

Arena Code WebDev — Consumer Product

Code

Arena Code WebDev rating for consumer-product and platform applications. Anonymous generated applications are compared through blind pairwise user votes.

Top models (higher is better)

ModelScore
Claude Opus 5.51705
GPT-6 Astra1702
GPT-6.1 Sol1685
Muse Wonder Full1661
Claude Fable 5.11661
GPT-6 Sol1649
Qwen3.8-Max1637
Claude Sonnet 5.51636
Claude Opus 51632
Muse Wonder1628
Qwen3.8 Max (0902)1620
DeepSeek V4.1 Flash 10171611
Kimi K31610
Muse Spark 1.31609
Kimi K3.1 Pro1607
Gemini 4 Argon1607
DeepSeek V4.1 Flash1588
Grok 4.61583
Hy4 Preview1578
MiMo-V2.6-Flash1574
Gemini 3.7 Flash1574
GPT-5.6 Sol1573
GLM-5.31572
Claude Fable 51571
Grok 4.71570
MiMo-V2.6-Pro1567
GLM-5.21564
DeepSeek-V4-Flash1563
DeepSeek V4 Pro 08131561
Qwen3.8-Flash-Next1560
Qwen3.8 27B1559
GLM-5.3 Flash1558
Gemini 3.8 Flash1548
GPT-6 Luna1547
Mistral Large 4 Preview1544
Opus 4.71542
Opus 4.81541
MiniMax M3.1 Flash1537
Opus 4.61530
Muse Spark 1.11530
Loading Atlas data…