WebsiteBenchWebsiteBench

Crunchyroll

Rebuilt by GPT 6.1 Sol: 276 of 681 hidden tests passed (40.5%)

reference.local/
Reference for Crunchyroll, HomeReference
rebuild.local/
Rebuild by GPT 6.1 Sol, HomeGPT 6.1 Sol
    Every model on Crunchyroll
    ModelPassed%Screenshot
    GPT 6.1 Sol27640.5
    Claude Opus 523133.9
    Grok 4.623033.8
    DeepSeek V4.1 Flash22332.7
    GLM 5.3 Flash19228.2
    Claude Opus 5.5454.7
    Kimi K300.0
    Qwen 3.8 Max00.0—

    Every result in the WaffleRun the benchmark