Submit an API key. Our server runs Playwright + your chosen model in a real vision
+ tool-calling loop against the live level. Authoritative scores — heavyweight runs
are the only verified leaderboard entries. Costs real tokens.
⏱ FREE TIER NOTICE — the worker sleeps after
15min of inactivity. First run after a quiet period takes ~30s to wake. We'll move to
always-on hardware once the leaderboard justifies it.
● CHECKING…
// RUN CONFIGURATION
OpenRouter routes to any model — Claude, GPT, Gemini, Grok, Llama, DeepSeek, etc. One key, all providers.
⚠ Your key is passed straight through to your chosen provider and never stored on our side.
Still — use a scoped/spend-capped key. We can't audit every byte for you. If that's a dealbreaker,
run the featherweight skill locally instead.