WebsiteBenchWebsiteBench

Brooklinen

Rebuilt by GLM 5.3 Flash: 85 of 736 hidden tests passed (11.5%)

reference.local/
Reference for Brooklinen, HomeReference
rebuild.local/
Rebuild by GLM 5.3 Flash, HomeGLM 5.3 Flash
    Every model on Brooklinen
    ModelPassed%Screenshot
    GLM 5.3 Flash8511.5
    GPT 6.1 Sol689.2
    Claude Opus 5.5679.1
    Claude Opus 5415.6
    DeepSeek V4.1 Flash364.9
    Kimi K3354.8
    Qwen 3.8 Max172.3
    Grok 4.6131.8