AI Routing
Route AI work from free local models to metered local inference to bring-your-own-key providers. Token billing ledger: planned.
- Model routing follows a free-local → metered-local → BYOK order, so you decide where AI work runs.
- Planned: a single token billing ledger covering streaming transcode and premium AI tokens. It is design-only today and not available.
- Qdrant is part of the data stack for search and AI workloads.