WebsiteBenchWebsiteBench

Target

Rebuilt by Claude Opus 5.5: 117 of 537 hidden tests passed (21.8%)

reference.local/
Reference for Target, HomeReference
rebuild.local/
Rebuild by Claude Opus 5.5, HomeClaude Opus 5.5
    Every model on Target
    ModelPassed%Screenshot
    Claude Opus 5.511721.8
    DeepSeek V4.1 Flash10619.7
    Kimi K39918.4
    GPT 6.1 Sol9818.2
    GLM 5.3 Flash7514.0
    Grok 4.67514.0
    Qwen 3.8 Max7413.8
    Claude Opus 5Error—

    Every result in the WaffleRun the benchmark