// the find
openmeterio/openmeter
Metering and Billing for AI, API and DevOps. Collect and aggregate millions of usage events in real-time and enable usage-based billing.
OpenMeter is a usage metering and billing engine targeting AI/API products — you send CloudEvents, define aggregation meters, and it handles the full billing lifecycle from usage tracking to invoicing. The stack is ClickHouse for real-time aggregation, Kafka for ingestion, and Postgres for billing state. Aimed at SaaS and AI companies that need usage-based pricing without building the plumbing themselves.
CloudEvents as the ingestion format is the right call — it's a real standard, not a bespoke schema you'll regret later. The separation between ClickHouse (hot aggregation) and Postgres (billing state) is architecturally sound and will actually scale. API-first design via TypeSpec generating both OpenAPI spec and SDKs means the clients stay in sync. First-class LLM token cost tracking with model-specific pricing is genuinely useful and ahead of most competitors.
Self-hosting requires Kafka, ClickHouse, and Postgres running together — the operational burden is heavy for a small team and the quickstart glosses over this. The Nix + direnv dev requirement will block contributors who aren't already in that ecosystem. Entitlement and subscription models are complex enough that mid-cycle changes and proration edge cases are almost certainly where bugs live, and the test surface for those paths isn't visible from the repo surface. No clear migration path documented if you start on cloud and want to move self-hosted, which is a real concern for cost-sensitive teams.