WebsiteBenchWebsiteBench

IMDb

Rebuilt by Kimi K3: 183 of 509 hidden tests passed (36.0%)

reference.local/
Reference for IMDb, HomeReference
rebuild.local/
Rebuild by Kimi K3, HomeKimi K3
    Every model on IMDb
    ModelPassed%Screenshot
    Kimi K318336.0
    Qwen 3.8 Max18235.8
    Claude Opus 5.58316.3
    DeepSeek V4.1 Flash7915.5
    GPT 6.1 Sol7915.5
    Grok 4.67915.5
    Claude Opus 57715.1
    GLM 5.3 Flash6913.6