Fallbakit Platform
Usage Stats
Read request logs, routing split, latency, cloud spend, and estimated local savings.

Requests
The Requests page shows recent route decisions with request time, selected backend, model mapping, latency, token counts, cost, savings, and error code.
Use this page to answer:
- Did the request route locally or to cloud fallback?
- Which model was requested and which model actually served?
- Was fallback caused by offline, slow, or degraded local inference?
Usage
The Usage page aggregates traffic over time. It helps you monitor:
- Local versus cloud request split.
- Estimated cloud spend.
- Estimated local savings.
- Average latency.
Errors
Errors such as application_not_allowed, provider failures, or backend errors appear in the request log. Check the selected application, allowlist entries, and fallback credentials before changing SDK code.