Know what AI costs. Control what it spends.

A transaction is one write to our metering API. Every time you call revenium.track(), that's one transaction.

Streaming responses: 1 transaction per completion (not per chunk)
Multi-step chains: 1 transaction per step you explicitly track
Batch sends: each event in the batch = 1 transaction

You control what you track. We count what you send.

Traces are for debugging. Transactions are for economics.

Every AI call has a cost. We treat it as a financial event that can be to attributed, audited, and governed. That's why we call it a transaction.

Dashboards, charts, API queries, and data exports are unlimited and free on all plans. We only meter writes (ingestion).

We bill overages monthly at $0.10 per 1,000 transactions. You can set alerts to stay ahead of usage.

No. We charge month-to-month. Cancel anytime. We earn your business every billing cycle.

Those are observability tools that show what happened.

Revenium is an economic control system. We attribute cost to business outcomes (customers, features, margin) and enforce limits before losses occur.

We often run alongside observability, not instead of it.

Stripe and Metronome monetize usage (billing). Revenium governs AI economics upstream. We meter and attribute cost before the invoice, so you know your margin before you bill.

OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Google Vertex AI, and any custom model via our SDK. See Integrations for the full list.

Yes. You can start on the free Observe plan and upgrade to Optimize anytime. Your historical data (within retention) carries over.

We work with early-stage teams on a case-by-case basis. Contact Sales to apply.