Open core · Go + PostgreSQL · self-host or use our cloud
ABAXUS is an open-core metering and billing engine. The core is free and runs inside your own cluster. ABAXUS Cloud is the same engine hosted and operated by us, with spike control and an agent that resolves metering errors before they become disputed invoices.
Why open core, why now
In twelve months every independent metering vendor became part of a payments or API platform. The software you pick today can change owners tomorrow. What should not change is who holds your usage data and the rules that turn it into invoices.
ABAXUS keeps both in your infrastructure, in an open core you can read, fork and run without us.
One engine, two ways to run it
Self-host the core and your usage data stays in your cluster. Choose ABAXUS Cloud and we host and operate the same engine for you, so there is nothing to run. Either way you can read the code and leave.
Everything you need to meter, price and invoice correctly. No event caps, no customer caps, no phone-home.
The same open core, operated by the people who wrote it, plus the parts that are hard to build and painful to run at 3 a.m.
Cloud · Spike control
A retry storm, a runaway AI agent or a late backfill hits a hosted billing API and you get 429s, dropped events and an invoice nobody can defend. ABAXUS absorbs the burst and decides what is billable before it reaches the ledger.
Every event is accepted and queued in PostgreSQL. Bursts slow the meter down; they never lose data.
Set baselines per meter and customer. One noisy tenant can't starve the rest.
Usage far outside a customer's pattern is held for review before it is invoiced, not refunded after.
Warn or cap customers at thresholds you define, so a surprise bill never reaches their CFO.
Cloud · Agentic dispute resolution
Duplicate events, late events, clock skew and meter-version mismatches turn month-end into an engineering fire drill. The ABAXUS agent investigates them against your own ledger and proposes a fix. A human approves. Nothing is ever overwritten.
Continuous checks between dashboards, aggregates and draft invoices flag mismatches before the invoice goes out.
The agent follows the SHA-256 trail from invoice line to meter version, plan and source events, and names the root cause.
An adjustment entry plus a plain-language explanation your support team can send to the customer as-is.
You approve or reject. Every decision lands in the audit trail, and the agent works against your own ledger.
case DSP-0182 · customer acme-labs · opened by support claim "October api_calls invoice is ~4% higher than our logs" found 2,904 events with identical idempotency keys, different timestamps cause client clock skew of +38s caused retries to land in a new window impact INV-3107 line 2 overstated by 2,904 units fix ADJ-0533 −2,904 units · meter api_calls.v3 · dedupe window → 120s status awaiting approval · evidence bundle attached
For the engineers evaluating this
{
"idempotency_key": "evt_01J…",
"customer_id": "acme-labs",
"metric": "api_calls",
"timestamp": "2026-10-31T09:14:02Z",
"properties": { "region": "eu-1" }
}
→ 202 Accepted{
"customer_id": "acme-labs",
"metric": "api_calls",
"meter_version": 3,
"consistency": "exact",
"quantity": "1204331.0000000000"
}{
"invoice_id": "INV-3107",
"lines": [ {
"metric": "api_calls",
"trace": "sha256:9f2c…e1"
} ]
}Go, PostgreSQL 16, Helm and Prometheus by default. The architecture is modular: storage and messaging are adapter interfaces, so the defaults need no Kafka, Redis or Elasticsearch, and you can swap an adapter when your platform requires it. Read the 8-step walkthrough
| Hosted billing SaaS | ABAXUS Core | ABAXUS Cloud | |
|---|---|---|---|
| Where usage events live | Vendor's cloud | Your cluster | ABAXUS Cloud |
| Who operates it | Vendor | You | ABAXUS |
| Who defines the deployment | Vendor | You: adapters and topology in your VPC | ABAXUS |
| Traffic spikes | API rate limits, retries | Durable queue, no drops | + burst budgets, anomaly holds, caps |
| Invoice disputes | Support ticket to vendor | SHA-256 trace, you investigate | + agent investigates and proposes |
| Exit cost if the vendor is acquired | Migration project | None, you hold the code | Same open core: move to self-hosted Core |
| PHI / residency exposure | Crosses a boundary, needs a BAA | Stays in your environment | Hosted by ABAXUS: choose Core if it must stay in-house |
Evaluating a specific vendor? See ABAXUS vs Lago, Metronome, Orb and Stripe Billing.
B2B SaaS from $5M to $100M ARR, usage-based or hybrid pricing, six or more engineers.
API calls, tokens, build minutes, compute. Event volume past vendor limits, per-transaction fees climbing with ARR, multi-region aggregates you reconcile by hand.
Patients, scans, records, claims. When PHI sits inside usage events, a third-party billing API becomes a compliance boundary, a BAA negotiation and a residency question.
Pricing
$0 forever
Self-hosted. Full metering, pricing, invoicing and audit trail. Community support on GitHub.
Deploy open source$4.2K /yr early adopter
$12K/yr standard at GA
Custom
Higher volumes, multi-region, custom retention and compliance reviews. Bi-weekly architecture calls.
Talk to an engineerFull tier limits and comparison on the pricing page.
Yes. Metering, pricing, invoicing and the audit trail are in the core with no caps. Cloud adds operations, spike control, the dispute agent and prebuilt connectors.
On ABAXUS Cloud your events are sent to the ABAXUS-operated service, which runs the metering pipeline, the agent and your ledger. If usage data must stay inside your own environment, run the self-hosted Core.
No. It proposes append-only adjustment entries with an evidence bundle. A person approves or rejects each one, and both outcomes are recorded in the audit trail.
You keep running the core. Your ledger, meters, plans and invoice history stay in your database. You lose the managed operations and the Cloud-only features.
To be confirmed before the core is published. We will announce the license here and on GitHub.
Yes. Growth includes migration help, including importing invoice history so past invoices stay explainable after the switch.
Start self-hosted in one command, or book a 30-minute architecture review with the engineer who built the ingestion pipeline.
Not ready to talk? Take the Billing Readiness Assessment, no sales call.