One gateway, all your model spend in view.
OpenRouter routes one API to hundreds of models across dozens of providers, and returns the exact cost of every call it makes. Obsivara ingests that — each request traced with the model and provider it landed on, tokens, latency, and dollars — so cross-model spend and fallback behavior stop being a single OpenRouter line item. Send OpenRouter's usage data via the Obsivara SDK, OpenTelemetry, or a usage event, out of the request path.
OpenRouter, fully observable.
Cost per model and provider
OpenRouter returns per-call cost; Obsivara rolls it up by the model and upstream provider each request actually hit, so you see exactly where spend goes across your whole gateway.
Cross-model comparison
With many models behind one API, Obsivara compares them on cost, latency, and reliability for the same workload — so model choice becomes an evidence-based decision.
Routing and fallback visibility
See when OpenRouter fell back to a different model or provider, and what that fallback cost in dollars and latency.
Reliability monitoring
Error rates, timeouts, and latency per model and provider, with health scoring and anomaly alerts.
Three steps. No code changes.
Send your OpenRouter usage
Instrument your app with the Obsivara SDK or an OpenTelemetry exporter, or POST one usage event per request — OpenRouter returns per-call cost and the model used, which Obsivara ingests directly. Your app keeps calling OpenRouter directly, out-of-band, with no proxy and no added latency.
Obsivara maps your usage
Within minutes you get a live inventory of every model and provider you route to, the workflows behind them, and what each costs.
Set budgets and alerts
Spend thresholds, latency SLOs, and error-rate alerts per model or workflow. Obsivara watches from there.
Signals tracked out of the box.
- →Cost per model and upstream provider
- →Tokens per request (input / output)
- →Model and provider routing per call
- →Fallback events and their cost
- →Latency and error rates by model
- →Health score and spend anomalies
Common questions.
Does Obsivara use OpenRouter's per-call cost data?
Yes. OpenRouter returns the exact cost and model for each request; Obsivara ingests that via usage events, the SDK, or OTLP, so spend is accurate without re-pricing.
Can I see spend broken down by the model OpenRouter actually used?
Yes. Every call is attributed to the model and upstream provider it landed on, so a single gateway bill becomes a per-model, per-provider breakdown.
Does it capture fallbacks between models?
Yes. When OpenRouter routes or falls back to a different model, Obsivara records the switch and what it cost in dollars and latency.