How it works
Every engine models cost from representative published-list prices — the rates vendors publish, not your negotiated, contracted, or private rates, and not your actual invoice. Figures are scenario estimates to reason with, not a quote or a bill.
Each rate carries a source link and a verified date (see Sources & Freshness). AWS rates are for a single reference region (us-east-1). When a rate is stale, or a resource isn't in the verified table, the engine fails closed — it suppresses the total rather than guessing.
Every lever is gated against the real billable meter. If an optimization doesn't change what you are actually charged — scan reduction under capacity billing, removing required high availability, quantizing a precision-locked workload, shortening retention within a billable tier — it returns $0 or a negative net, and says so. There is no guarantee of any savings; results depend entirely on the assumptions you enter.
The math runs entirely in your browser. Quote-only, private, and contracted prices are not modeled. Always verify against your own invoice and the provider's current pricing before acting.