Opsmeter.io
AI Cost & Inference Control
Integration docs
Implementation guide for AI cost control
Ship telemetry fast, keep attribution stable, and move from ingest to governance without changing your app network path.
Recommended path
Start with direct ingest, validate one stable payload shape, then harden budgets, retries, and incident workflows.
What this page answers
- How to keep the app network path unchanged while sending telemetry.
- Which metadata fields matter most for request-level attribution.
- How to preserve operational visibility without adding a proxy.
Implementation rhythm
5 minfirst ingest path
No proxyapp path unchanged
One schemaacross providers
Use quickstart for initial payloads, then operations docs to harden production workflows.
No-proxy telemetry
No-proxy LLM telemetry: track cost after each call
Opsmeter.io is production-ready without an SDK wrapper. Keep your provider call path unchanged, map provider usage fields, and send telemetry asynchronously with a short timeout.
Implementation model
- Call provider SDK/API and read usage metadata.
- Normalize payload with endpointTag, promptVersion, userId, and token fields.
- Send ingest asynchronously with timeout + swallow to protect request latency.
What is available now
- Endpoint, promptVersion, and user/tenant attribution.
- Budget warning and exceeded workflows.
- Export, retention policy, and compare workflows.
SDK roadmap
- Automatic wrappers for common frameworks.
- Runtime clamp/fallback/queue enforcement patterns.
- Policy contracts for machine-readable guardrail actions.
SDK package links and current support live in Quickstart → SDK packages.