WebsiteBenchWebsiteBench

Autotrader

Rebuilt by Claude Opus 5: 274 of 487 hidden tests passed (56.3%)

reference.local/
Reference for Autotrader, HomeReference
rebuild.local/
Rebuild by Claude Opus 5, HomeClaude Opus 5
    Every model on Autotrader
    ModelPassed%Screenshot
    Claude Opus 527456.3
    GLM 5.3 Flash25853.0
    GPT 6.1 Sol22746.6
    Claude Opus 5.520441.9
    Grok 4.68818.1
    Kimi K37214.8
    Qwen 3.8 Max193.9
    DeepSeek V4.1 Flash40.8