Does handing the choice to a router beat naming a model?
A router executes a routing policy per request. Vitna evaluates whether that policy actually beats a fixed model on your workload. Because routers have no published list price, the economics can only come from your own deployment telemetry and evals. The scenarios below are illustrative: no router was run for them, and they name no product.
Illustrative. No router was run: the counts, fees, success rates and latencies are written to show the arithmetic, and the scenarios name no product.
illustrative/summarizationA per-request-fee router yields 10.0% savings, which does not clear the 20.0% threshold required to justify routing complexity and latency risk.
| Check | Required threshold | Observed | Result |
|---|---|---|---|
| quality floor | 85.0% | 90.0% | Passed |
| critical failures | 0 veto failures | 0 failures | Passed |
| economic threshold | 20.0% | +10.0% | Failed |
| observation threshold | 20 units | 1000 units | Passed |
| Metric | Incumbent | Candidate router | Difference |
|---|---|---|---|
| Raw unit cost | $0.0200 | $0.0180 | +10.0% |
| Task success rate | 90.0% | 90.0% | +0.0% |
| Successful unit cost | $0.0222 | $0.0200 | +10.0% |
| Latency p90 | 800ms | 1100ms | +300ms |
retrospective_deployment_observation