Borrow, rent, or build?
Pick your workload and see the honest monthly cost of each route — build (own the GPUs), rent GPUs, pay per-token via API, or reserve provisioned throughput. Same deterministic engine as the Capacity Planner, distilled to the decision.
Costs carry a ±50% accuracy band from the engine's back-tests. Archetypes are ranked on cost only — never on reliability or latency, which we don't measure yet.
Want the full breakdown?
The Capacity Planner shows the sizing math, per-line self-host costs (capex/power/colo/staff), break-even utilization, and the token budget behind these numbers.
Open the Capacity Planner“Rent” spans marketplace and neocloud at the cheapest live rate; hyperscaler on-demand and dedicated contract pricing are separate rental classes on the roadmap. No route is ranked on reliability — we don't measure it yet.