If your product charges by the token, the API call, or the minute of compute instead of a flat monthly fee, Stripe's default billing tools stop being enough pretty quickly. Usage-based pricing means you need real-time metering, credit systems, proration, and invoices that actually match what a customer consumed — and building that in-house is a multi-month engineering project most teams would rather skip. A specific category of "usage-based billing platform" has grown up around this exact problem, mostly serving AI and infrastructure companies whose pricing looks nothing like a normal SaaS subscription. Here are six real platforms in that space, from developer-first metering engines to fully open-source options.
The short version: if you want the most battle-tested option and don't mind an enterprise sales process, Orb is the category leader, trusted by companies like Vercel and Replit. If you want transparent self-serve pricing with real dollar figures on the page, Amberflo and Flexprice both publish theirs. If vendor lock-in worries you, Flexprice and Meteroid are genuinely open-source and self-hostable. And if you'd rather not run billing infrastructure at all, Dodo Payments and Kelviq act as merchant of record and take the tax compliance off your plate entirely.
Orb — the category leader for complex, at-scale pricing
Orb lets you define pricing metrics in SQL, which sounds like a small detail until you've tried to model a genuinely complex usage formula in a no-code rule builder and hit a wall. It processes over a million billing events per second and bundles finance workflows — ASC 606 revenue recognition, accounts receivable, dunning — that most competitors leave to a separate tool. It's used by infrastructure-heavy companies like Vercel, Replit and LaunchDarkly.
Good for: SaaS and AI/infrastructure companies with genuinely complex or metered pricing, and finance teams that need revenue recognition automation built in. Limits: there's no public pricing — you go through an enterprise sales process to get a number — and the SQL-based configuration means you'll want engineering time to set it up, not just a finance person with a spreadsheet.
Amberflo — real-time metering with actual published prices
Amberflo tracks usage in real time rather than reconciling it in a batch job later, and it specifically tracks AI/API costs across 1,500+ LLM models through a built-in cost gateway, with budget alerts designed to catch runaway spend before the invoice arrives. Pricing starts at $99/month (Startup) after a 30-day free trial, jumping to $599-$1,499/month at the Growth tier.
Good for: AI-native companies that specifically need LLM cost visibility alongside customer billing, and teams that want a real number on the pricing page instead of a demo request. Limits: it's competing directly in an already crowded field against Orb and Stripe Billing, the AI-cost gateway is the newest and least proven part of the platform, and the jump from $99 to $599+ at Growth is steep.
Flexprice — open-source, with a genuinely usable free tier
Flexprice is actually open source — self-hostable, not just a SaaS product with a free plan bolted on — and it's built natively for AI-style credit-based pricing, not just flat usage tiers. It claims sub-60ms metering at 10,000+ events per second. The free tier covers 100K events/month up to a $100K cumulative revenue cap; Build is $500/month (1M events, $250K cap) and Scale is $1,000/month (5M events, $1.2M cap).
Good for: engineering teams at AI-native or usage-based SaaS companies who want to avoid vendor lock-in and are comfortable running (or at least evaluating) open-source infrastructure. Limits: free and Build tiers are capped by cumulative revenue as well as event volume, so fast-growing companies will outgrow them sooner than the sticker price suggests, and the ecosystem is younger and smaller than Stripe Billing or Orb's.
Meteroid — fully open-source, no cloud lock-in required
Meteroid is AGPL-3.0 open source and free to self-host indefinitely, built in Rust, and API-first, with support for complex matrix and hybrid pricing models. Managed cloud tiers exist for teams that don't want to run it themselves, though that pricing isn't published.
Good for: technical SaaS teams who specifically want to own their billing infrastructure rather than depend on a vendor, and who are comfortable with the operational tradeoff that comes with it. Limits: at roughly 785 GitHub stars, it's a smaller and younger community than the established players; self-hosting means you own uptime, scaling and security patching yourself; and the AGPL license has copyleft implications worth a legal read before you adopt it.
Dodo Payments — merchant of record for AI-first and indie SaaS
Dodo Payments acts as merchant of record, meaning it assumes the VAT/GST tax compliance burden across 220+ countries instead of leaving it to you, and it supports credit-based, usage-based, subscription and hybrid billing in one platform, plus license-key management for digital products. Pricing is 4% + $0.40 per transaction with no monthly fee, though extra fees apply for BNPL/PayPal (+3%), international cards (+1.5%), and chargebacks ($30).
Good for: indie hackers, micro-SaaS founders and AI-first companies selling API or credit-based products globally who don't want to build tax compliance themselves. Limits: 4% + $0.40 per transaction is meaningfully more expensive than raw payment processing, the extra fees stack up for international or BNPL transactions, and being merchant of record means less direct control over checkout and payout timing than a raw processor gives you.
Kelviq — usage-based billing plus global tax compliance
Kelviq combines usage-based pricing built specifically for AI metrics (tokens, compute units) with merchant-of-record tax compliance across 100+ countries, and ships typed SDKs across TypeScript, React, Node and Python along with license keys and feature entitlements in one system.
Good for: SaaS and AI companies selling internationally that want usage-based billing and tax compliance handled by the same vendor. Limits: pricing isn't public — you have to book a demo to get a real number — and there's no disclosed launch date or public track record yet, which makes it harder to comparison-shop against Paddle or Stripe.
Side-by-side comparison
| Tool | Price | Open source? | Best for |
|---|---|---|---|
| Orb | Custom, enterprise sales | No | Complex metered pricing at scale |
| Amberflo | $99–$1,499+/mo | No | Real-time AI/LLM cost tracking |
| Flexprice | Free–$1,000+/mo | Yes | Self-hostable, credit-based AI pricing |
| Meteroid | Free (self-host) + custom cloud | Yes (AGPL-3.0) | Full ownership of billing infra |
| Dodo Payments | 4% + $0.40/transaction | No | Merchant-of-record, indie/AI SaaS |
| Kelviq | Custom, demo required | No | Usage billing + global tax in one |
None of these six tools solve exactly the same problem the same way, which is really the point: usage-based billing isn't a solved, commodity category yet the way flat-rate subscription billing is. If your pricing is genuinely complex and you have engineering time, Orb or an open-source option like Flexprice or Meteroid make sense. If you'd rather hand off tax and compliance entirely, Dodo Payments or Kelviq are built for exactly that. Whichever you pick, the real test is whether the platform's metering model matches how your product actually generates cost — not just how clean the pricing page looks.