Lago is an open-source billing platform that meters usage and invoices at cycle end, with premium features gated behind paid Business and Enterprise tiers. Credyt is real-time billing infrastructure that debits prepaid wallets before usage occurs. This article compares their architectures, total cost of ownership, and when each fits.
At a glance
Lago: An open-source (AGPLv3) billing platform for usage-based and subscription billing. Series A funded with 9,500+ GitHub stars and 183 releases through v1.45.1 in April 2026 (Lago GitHub). The core is free to self-host; customer portal, credit notes, automatic dunning, real-time balance observability, and AI-native features sit behind paid cloud tiers documented on Lago's premium features page.
Credyt: Real-time billing infrastructure for AI products where infrastructure costs hit in real time. Wallets are first-class primitives, usage debits happen atomically on event ingestion, and the platform can query wallet state to authorize actions before costs are incurred. Proprietary, cloud-only, published pricing at $1 per Monthly Active Wallet.
What Lago does
Lago ingests usage events in real time and supports subscriptions, usage-based pricing, hybrid plans with credit allowances, coupons, and entitlements. The platform generates invoices at cycle end, handles payment orchestration through native connectors for Stripe, Adyen, and GoCardless, and exposes a REST API plus a hosted dashboard for pricing configuration. Named customers include Mistral AI, PayPal, Groq, Synthesia, Arcee AI, and 1NCE: a mix of AI infrastructure, fintech, and enterprise. The self-hosted edition ships as a Docker Compose stack; the managed cloud offers Business and Enterprise tiers with additional premium features.
Where Credyt differs
Credyt debits wallets atomically on each event; Lago meters in real time but applies credits at invoice finalization, so the authoritative balance moves at cycle end rather than per request. Three architectural axes separate the two platforms: billing timing, pre-usage authorization, and the scope of what ships included in the base tier.
For teams billing fixed subscriptions with end-of-cycle reconciliation, Lago's model fits cleanly. For real-time billing for AI products where a single inference call can cost $1 or more and customer behavior is spiky, waiting until cycle end to reconcile consumption leaves cost exposure uncontrolled.
The second axis is authorization. Lago's documented model is observe-only. The platform can react to consumption but cannot query a wallet and block a specific action before it runs, because the authoritative balance is not updated in real time. Credyt exposes a Wallet API the platform can call before each action; if balance is sufficient, the action proceeds and Credyt charges and debits immediately.
The third axis is scope of what ships included. Lago's AGPLv3 core covers metering and invoicing. The customer portal, credit notes and refunds, automatic dunning, tax integrations, accounting connectors, and real-time ongoing balance are paid features (Lago pricing; Lago premium features: customer portal). Credyt bundles the portal, wallets, hybrid billing, and real-time balance into core at a published $1 per Monthly Active Wallet with the first 10 wallets free.
Technical comparison
The most consequential architectural gap is authorization: Credyt exposes a Wallet API the platform can call before each action; Lago has no equivalent primitive to block usage before it occurs. The table below compares both platforms across the dimensions that most affect billing architecture decisions.
| Dimension | Credyt | Lago |
|---|---|---|
| Billing model | Hybrid (real-time wallet debit + subscription entitlements) | Invoice (real-time metering, end-of-cycle invoicing with credit-on-invoice) |
| Usage authorization | Pre-usage (platform queries Wallet API before the action) | Post-usage observe-only (no primitive to block before action) |
| Usage tracking | Atomic per-event debit | Metering + credits applied at invoice finalization |
| Wallet architecture | First-class primitive; atomic balance checks | Add-on; up to 5 wallets per customer; credits reduce invoice totals |
| Real-time balance visibility | Core; atomic on each event | Premium only; 5-minute refresh on ongoing_balance_cents |
| Multi-asset support | Native (USD, tokens, GPU hours, custom units) | Fiat with display labels |
| Customer portal | Drop-in, branded, pre-authenticated (core) | Premium feature |
| Refunds and balance adjustments | Adjustments API (core) | Credit notes module (premium) |
| Open source | No | Partial (AGPLv3 core; premium features are commercial) |
| Deployment | Cloud only | Self-hosted (Docker) or Lago Cloud |
Sources: Lago docs, Lago pricing, Credyt product docs, accessed April 2026.
Pricing at the canonical scenario
At 100 customers with $20 average monthly spend, Credyt costs $90 per month plus PSP fees. Lago's cloud pricing is not publicly available and requires a sales conversation; self-hosted OSS is free to license but carries infrastructure and engineering costs.
The canonical comparison scenario for reference: 100 customers, $20 average monthly spend per customer, 1,000 events per customer per month, 5 platform seats. Monthly revenue through the platform is $2,000; total monthly events are 100,000.
| Component | Credyt | Lago Cloud | Lago Self-hosted OSS |
|---|---|---|---|
| Platform / base fee | $0 | Not publicly available | $0 (AGPLv3 license) |
| Per-customer fees | $90 ($1 × 90 MAW; first 10 free) | Not publicly available | $0 |
| Seat fees | $0 | Not publicly available | $0 |
| Event fees | $0 (first 1M events free) | Not publicly available | $0 |
| Revenue % overage | $0 | Not publicly available | $0 |
| Payment processing | Pass-through (PSP rate) | Pass-through | Pass-through |
| Infrastructure | Managed | Managed | Postgres + Redis + compute for 5+ app services + object storage + observability |
| Engineering effort | SDK install, event wiring, pricing config | SDK install, event wiring, pricing config, webhook coordination, invoice lifecycle handling | All of cloud's integration work plus deployment, upgrades, incident response, and building or paying for premium features not in OSS |
| Premium features | Portal, wallets, real-time balance, hybrid billing, metering, analytics all included | Varies by tier; customer portal, credit notes, dunning, tax integrations all paid | None included; require paid tier or custom build |
| Total monthly | $90 + PSP | Not publicly available; requires sales | Infra + engineering time; premium feature parity requires paid tier or custom build |
Lago cloud pricing is not published. During Lago's April 2024 Series A announcement thread on HN, commenter plantain referenced a $3,000-per-month starting cloud tier; commenter lucw reported $1,500 per month as the minimum premium tier they needed to access the customer portal (HN: Lago banks $22M, April 2024). Current cloud pricing may differ and is not disclosed publicly.
When to choose Lago
Lago is a strong fit in specific situations:
- Your team wants full code transparency and the ability to audit every line of the billing logic.
- Compliance or data residency requirements mandate self-hosting.
- You have dedicated engineering to deploy, operate, and upgrade the Postgres, Redis, Rails, Sidekiq, and Gotenberg stack on a continuous release cadence.
- Near-term scope does not require a customer portal, real-time balance observability, credit notes, automatic dunning, or tax integrations.
- Event volumes and reliability requirements justify the fixed operational cost of a production self-hosted deployment.
- You are an enterprise buyer comfortable engaging on the Enterprise tier for premium features with an assigned solutions engineer.
When to choose Credyt
Credyt is a stronger fit in a narrower but common set of cases:
- You are building an AI product that bills for tokens, API calls, or compute with real-time costs.
- You need pre-usage authorization: checking wallet state and blocking an action before expensive infrastructure runs.
- Customers expect a self-serve portal with live balance, usage history, and top-ups without waiting on sales.
- You want published pricing and no procurement cycle.
- You need to ship billing in days rather than weeks.
- You have no appetite for operating Postgres, Redis, Rails, Sidekiq, and PDF rendering as part of your billing stack.
How do the architectures differ?
Credyt and Lago diverge most on three dimensions: wallet debit timing (atomic per-event in Credyt vs invoice-finalization in Lago), customer portal availability (core in Credyt, premium-only in Lago), and integration posture (Credyt extends an existing PSP stack; Lago replaces the billing layer).
Wallet mechanics: real-time debit vs invoice-finalization debit
Lago: Lago supports prepaid credit wallets, but the drawdown model is not a per-event debit. The authoritative balance, balance_cents, updates when an invoice is finalized. Lago announced "real-time burndown" in July 2024 with messaging about instantaneous credit deduction; the docs clarify that the corresponding ongoing_balance_cents is an estimate, refreshes every 5 minutes, and is premium-only (Lago wallet overview). OSS users do not see the ongoing estimate at all; they see the invoiced balance that moves at cycle end. The model is observe-only.
Credyt: The wallet is debited atomically on event ingestion. The platform queries Credyt's Wallet API before an action; if balance is sufficient, the action proceeds and Credyt charges and debits immediately. Atomic balance checks prevent credit depletion races under concurrent requests.
For AI products where a single inference call can cost $1 or more and customer behavior is unpredictable, an invoice-finalization debit and a 5-minute-refresh estimate both discover exposure rather than prevent it. An atomic per-event debit with pre-usage authorization lets the platform authorize before the cost is incurred. Both models are legitimate; they fit different use cases. This is a reason AI companies look for alternatives to Lago when per-request costs are high and revenue must clear before cost is locked in. Our notes on Stripe metered billing issues cover adjacent terrain on post-paid billing in AI products.
Customer experience: drop-in portal vs premium feature
Lago: The customer portal is a premium feature. OSS users either pay for the cloud tier or build their own. Building a minimum-viable customer billing portal (live balance, invoice list, top-up button, auto-recharge configuration) takes around two weeks of engineering in Credyt's own experience building these, even with AI coding tools, and that is before branding, testing, and edge-case handling. HN commenter lucw put the alternative cost in concrete terms: "I couldn't do without the customer portal which is a premium feature, and premium was at least $1,500 USD/month" (HN: Lago banks $22M, April 2024).
Credyt: A drop-in, branded, pre-authenticated portal is part of core. Customers see live balance and usage history. They set their own auto top-up thresholds and top up via Stripe Checkout without contacting support.
Integration model: extend vs replace
Lago: A full billing platform. It replaces or absorbs most of the billing stack: invoice generation, PDF rendering, webhook coordination, payment provider wiring, dunning.
Credyt: Extends PSP and subscription tools. Works alongside Stripe, downstream billing systems, and ledger tools rather than replacing them.
Teams with an existing PSP and invoice system often want to add real-time control without migrating the whole stack. Credyt fits the extension path; Lago fits the replacement path.
Bottom line
The Credyt vs Lago decision comes down to two questions. Does your team want to own the stack, and do the features you will grow into sit on the free side of Lago's OSS-vs-premium line? Lago's OSS core is genuinely free to self-host for teams ready to operate the infrastructure; the customer portal, real-time balance observability, credit notes, and automatic dunning are paid features. Credyt bundles the portal, wallets, real-time metering, and pre-usage authorization into core at a published $1 per Monthly Active Wallet.
Frequently asked questions
What features are in Lago's free open-source version?
The AGPLv3 core includes real-time event metering, subscriptions, usage-based and hybrid pricing, coupons, entitlements, and core wallet mechanics with credits applied at invoice finalization. It does not include the customer portal, credit notes and refunds, automatic dunning, real-time ongoing balance, or tax, CRM, and accounting integrations, which Lago documents as premium features or add-ons at getlago.com/premium-features/customer-portal and related pages.
Does Lago support real-time wallet drawdown for usage-based billing?
Partially. Lago supports prepaid credit wallets, but the authoritative balance updates at invoice finalization rather than on each event. The ongoing_balance_cents that shows a real-time estimate of consumption is premium-only, refreshes every 5 minutes, and is observe-only; the platform cannot query it to block an action before the action runs (Lago wallet overview, accessed April 2026). For AI products where a single request can cost $1 or more, this architectural gap matters.
How much does Lago's managed cloud cost?
Lago does not publish cloud tier prices. Business and Enterprise both require sales contact as of April 2026 (Lago pricing). In April 2024, during Lago's Series A thread on HN, commenter plantain referenced a $3,000-per-month starting cloud tier and lucw reported $1,500 per month as the minimum premium tier needed for the customer portal (HN: Lago banks $22M). Current pricing may differ and is not disclosed publicly.
Does running Lago self-hosted actually cost nothing?
The AGPLv3 license is free. Operating Lago in production requires a managed Postgres with backups, a managed Redis, compute for the Rails API, Sidekiq worker, scheduler, front-end, and Gotenberg PDF service, object storage, observability, and ongoing engineering time for the 183-plus releases shipped through April 2026 (Lago docker-compose.yml, v1.45.1).
Does Lago take a percentage of my revenue?
Lago has historically positioned itself as not taking a revenue cut (Launch HN: Lago, February 2023). Current cloud tier terms are not publicly published, so revenue-share structures cannot be confirmed either way from public sources.
Can I add Credyt without replacing my existing billing system?
Yes. Credyt extends PSP and subscription tools rather than replacing them. Teams that already have Stripe for payments and a downstream invoicing system typically add Credyt as a real-time control and wallet layer on top. See how AI companies adopt real-time billing without replacing their stack for the adoption patterns.
What are some alternatives to Lago for usage-based billing?
If you are looking for alternatives to Lago, two directions are worth evaluating depending on your constraints. Credyt, covered above, is wallet-first with real-time authorization and managed ops. Flexprice is another open-source billing platform with a similar self-hosted model and its own OSS-vs-premium split; our Credyt vs Flexprice comparison covers the specifics.