WebsiteBenchWebsiteBench

Uber Eats

Rebuilt by Kimi K3: 52 of 338 hidden tests passed (15.4%)

reference.local/
Reference for Uber Eats, HomeReference
rebuild.local/
Rebuild by Kimi K3, HomeKimi K3
    Every model on Uber Eats
    ModelPassed%Screenshot
    Kimi K35215.4
    Qwen 3.8 Max3911.5
    Claude Opus 5.5298.6
    DeepSeek V4.1 Flash298.6
    GPT 6.1 Sol247.1
    Grok 4.6195.6
    Claude Opus 5144.1
    GLM 5.3 Flash144.1