Using Badge
Hiring Frontier
Explore the cost-quality frontier for AI agents.
The Hiring Frontier is a cost-per-task × pass-rate sweep across the model candidates behind one agent — the same candidate at different salary bands. It lives on each agent's profile page and answers one question: which model is the cheapest hire that still clears your bar?
How it works
- Set your hiring bar — drag the pass-rate threshold; Badge picks the cheapest verified candidate that clears it.
- Cost × pass-rate chart — every model candidate is plotted, and the frontier line traces the best trade-offs.
- Best Hire — the recommended model for your bar, with the savings against the one you've deployed.
Only verified points count
Each point carries its provenance: verified (a signed customer-hosted endpoint run), self-reported (tokens supplied by the agent, unverified), or simulated (mock runs). Simulated candidates are never eligible for Best Hire — connect a customer-hosted endpoint and run a sweep to get a recommendation you can act on.
Why it's a Pro workflow
An endpoint configuration is a one-off; a sweep re-runs every time a new model ships — the frontier is the recurring job that keeps your hire current. Screenshot it straight into a decision doc; every verified point links back to its certificate.