Calculator
What would routing save you?
Pick your current model and volume. The calculator sends the confident share of calls to Levanto Sage 2 ($0.05 per million input tokens) and the rest to your current model.
Yes/no tasks. Prices checked 2026-11-03. Estimates, not a quote.
The share the cheaper model keeps is the number that decides everything, and it depends on your task. Measure it on your own samples with the free audit →
How it’s calculated
- Now: calls × (input tokens × input price + output tokens × output price).
- Hybrid: the cheaper model on every call, plus your current model on the share it doesn’t keep.
- The same formula the audit uses. See the method.
Calculator questions
Where does the coverage number come from?
From your data. Coverage is the share of calls the cheaper model answers with enough confidence to keep, at the agreement level you need. The free audit measures it on 25 of your real samples; the slider here is a guess until you run it.
Why does the hybrid still pay for the cheaper model on every call?
Routing needs the cheaper model’s answer and confidence first. Calls below the threshold then go to your current model as well, so you pay for both on that share.
What does this leave out?
Batch and cached-input discounts, reasoning tokens (add them to output tokens if you use them) and your engineering time. Prices are list prices.