Understanding the cost estimate

The point estimate is scaled down 20% and up 35% to give the low–high range you see. That spread isn't padding — it reflects real variance a pilot shop's actual quote will have (machine rate, material batch pricing, how efficiently your specific geometry nests or machines) that a parametric model can't pin down without a real quote in hand.

• Machine time — derived from your part's volume and the process's per-hour rate (print time for FDM/SLA, material-removal time for CNC).
• Material — part mass × the material's cost per kg, with a process-specific waste factor (support/purge for FDM, resin + support for SLA, stock removal for CNC).
• Setup, amortised — a one-time setup cost spread across your stated volume. This is why volume matters so much: the same tooling setup at 20 units vs. 2,000 units changes the per-unit number by an order of magnitude.
• Post-processing — support removal, wash/cure, or a named finish, per part.

If your brief stated a target unit cost, the run flags whether the estimate is within it and names the driver behind the gap when it isn't. If a blocking DFM issue is still open, manufacturer matching and the cost picture pause until it's resolved — a cost number for a part that can't yet be made as designed isn't a useful number.

Every run is capped at $4 of actual agent spend, with a live token meter showing exactly what was spent and where. The cost estimate above is about what your part would cost to manufacture — a separate thing from what the run itself costs to compute.