Payments AI Pricing: Examples & Companies

What is it

Payments AI Pricing is pricing for the AI-enabled billing and payment infrastructure that helps software companies meter usage, generate invoices, and collect revenue. These are the picks-and-shovels of usage-based pricing — the platforms many other companies in this corpus buy to run their own usage billing — so their own price cards are a meta-example of the models they enable.

The distinctive thing is how self-referential the pricing problem is. A vendor selling usage billing to AI and SaaS companies must itself pick a meter: a percentage of the revenue it processes (its success rides on the customer’s growth), a flat platform fee (predictable for both), a per-event throughput meter, or a hybrid. Each answer shifts risk differently between vendor and buyer, and the 14 companies here span nearly the full range.

Consolidation now shapes the category from above: Stripe Billing’s parent closed its acquisition of Metronome in January 2026 (press reported roughly $1B), and Zuora acquired Togai, whose page now reads “Togai by Zuora.” As usage pricing becomes the default for AI companies, the metering-and-rating layer becomes platform-level infrastructure worth owning — and a relationship that is hard to reverse, because rate cards, historical usage data, and billing histories create durable switching costs in both directions. The usage invoicing and billing cycles guide is the prerequisite read before any vendor conversation.

How it works

Billing platforms sit between a company’s product-usage data and its customers’ invoices: ingest raw usage events → apply rate cards → compute line items → generate and deliver invoices → reconcile with payment. The pricing question is which part of that loop becomes the meter — and there are four structural answers.

The crossover between the first two rows is the whole decision (the visualization above prices one $200K/mo biller on each): a percentage meter is cheap when billings are small and unbounded when they are large, and a flat platform fee is its exact mirror. That is why the suites hedge with a hybrid — Chargebee runs a free Starter to $250K cumulative billing, a $599/mo Performance plan to $100K/month, then 0.75% overage above the cap, capturing the small biller (free) and the large one (percentage) with one grid.

For the events-metered variant, Flexprice publishes a clean ladder — Free to 100K events, Build $500/mo (1M events, to $250K cumulative billings), Scale $1,000/mo (to $1.2M billings), then custom — so either throughput (events) or value (cumulative billings) can trip an upgrade.

Unit math: On Flexprice’s Build plan, a startup ingesting exactly 1M events while invoicing $200K cumulative pays the flat $500/mo — the meter is pre-bought in bulk. Cross either ceiling and the next $1,000/mo tier applies; nothing scales continuously in between.

Companies using this

Fourteen companies in the corpus are tagged to the payments use case — the billing, metering, and quote-to-revenue platforms other software companies buy to run their own usage-based pricing. The range is wide: fully open-source engines (Lago, Kill Bill, Flexprice), sales-gated enterprise platforms (m3ter, Metronome, Zenskar), and publicly priced self-serve products (Hyperline, Maxio, Alguna, Schematic).

Patterns observed

Across the 14 companies, four structural patterns recur — each a choice about how much of a customer’s success the vendor wants to share.

• The percentage meter rides on an expensive base. Where a percentage-of-billings fee looks small, it usually sits on top of something larger. Stripe Billing can keep its billing rate low precisely because it already earns the ~2.9% + 30¢ card-processing fee underneath — the billing line is a platform upsell on money it is already moving. A pure billing vendor with no processor beneath it cannot afford the same discount, which is why standalone percentages land higher.
• “No % of revenue” is the counter-positioning. The incumbents’ meter becomes the challenger’s sales argument: Zenskar made “no % of your billed revenue” explicit alongside its $15M Series A, pitching finance buyers who have watched a percentage line item grow with their own success.
• Open-core is the top of funnel. Lago (AGPLv3, ~9.8k GitHub stars, ~$829M invoices processed monthly for customers including Mistral AI, PayPal, and Groq) and Flexprice give away the full engine to self-hosters and monetize operations, SLAs, and enterprise features — engineers validate integration before any sales conversation starts.
• Seat-free is table stakes. Finance and engineering teams buying billing infrastructure will not pay per seat, so challengers advertise it as a feature: Alguna and Schematic both list unlimited users on every tier, alongside published $0-to-Growth ladders that let buyers self-qualify in minutes against the sales-gated norm.

Counterexamples & variants

The model that charges on entitlements, not volume. Schematic is the cleanest counterexample to the events-and-billings paradigm. It meters “monetized subscriptions” — the number of your end-customers actively billed through the platform — with a hard ceiling at 10 (Free) and 100 (Growth, $200/month). This is entitlement infrastructure more than billing infrastructure: you pay for how many customers you’ve turned on a billing relationship with, not the events they generate. At the cap the free tier literally blocks additional customers from checking out, making the upgrade trigger deterministic. The tradeoff inverts event-metered pricing — high-usage customers with few relationships are cheap, low-usage customers with many are expensive.

Kill Bill’s explicit rejection of the revenue-share model. Kill Bill’s pricing page is built around one argument, backed by a savings calculator: its flat Aviate software tiers and support contracts (a representative ~$50k–$75k/yr at $25M ARR) are set so cost compresses as a share of revenue as the customer scales, instead of expanding with their success. The tradeoff runs the other way for the vendor — its revenue does not grow with customers, limiting reinvestment without adding accounts. Self-hosting also demands an engineering team to operate billing, a cost the price list never shows.

Where the percentage model breaks. Its canonical failure mode is the high-revenue, low-complexity customer buying the platform for simple invoicing. A company with $10M monthly recurring revenue on a plain annual-subscription structure would pay roughly $70,000/month in Stripe Billing fees for a product doing almost no metering work in their case. Volume discounting behind a sales conversation is the usual patch, but the sticker shock is real — it is what drives the “migrate off percentage billing” case studies flat-fee challengers publish prominently.

The managed-cloud moat without an open-source escape. Metronome and m3ter are proprietary with no open-source alternative and no self-serve path — m3ter’s only entry point is a “Schedule a demo” form, Metronome’s self-serve stops at a free starter. Both serve the high end (Metronome’s customers include OpenAI, Anthropic, and NVIDIA) where switching cost is accepted as part of the deal. But starting adoption in a sales conversation skews toward large buyers and makes the tools invisible to the engineering-led early-stage companies that might otherwise grow into them. Togai sits in the same fully sales-gated bucket — no published rates at all.

What this means for buyers vs vendors

For buyers

Open every evaluation with one question — “what is your meter?” — because events, invoices, billing volume, and monetized subscriptions price the same workload incomparably. Then force every bidder onto one twelve-month sample of your own event volume and invoice-value distribution: a business with high event volume but low average invoice value (SaaS monthly billing) lands in a completely different place than one with the reverse (enterprise annual contracts) on an events-plus-billings model.

Negotiate the percentage line hardest — it is the only component that compounds with your own success, so a rate that looks trivial today is a growing tax on tomorrow’s revenue. Get a flat-fee walk-away number to anchor the negotiation, and pin data portability in writing: rate cards, customer billing history, payment methods on file, and historical invoices are the four switching-cost lock-ins, so specify the export format and timeline for each. Weight acquisition risk too — contracts signed when a vendor was independent may not survive a consolidation intact. The introduction to usage-based pricing sets context, and the tracking and metering usage events guide covers the event pipeline your engineering team must wire before any platform is useful.

For vendors

Two lessons carry across the cohort. A percentage-of-billing-volume meter is the highest-value model at scale, but only if you either sit on an embedded processor or hold very strong contract discipline at the top — large customers will renegotiate the compounding line every renewal. And the open-source or free-tier wedge is a real acquisition channel: it lets engineering teams validate integration before procurement ever starts, which is why the credible players in this cohort all keep one door open at $0.

If you add a second dimension beyond your primary value meter, pick events processed or invoice count over seats — it scales with actual platform usage, not headcount, and this buyer will reject per-seat billing outright. Finally, treat transparency as positioning: if your pitch says buyers deserve fair, visible pricing while your own page shows none, the technically literate buyers you sell to will read the contradiction. Whether you publish a rate or gate it is itself a stance, and design your entitlements deliberately with the usage invoicing and billing cycles guide before you commit to either.

Explore this theme in the knowledge graph →