Estimate your revenue before choosing a plan — try the free cost calculator →
What teams are saying
“Our billing backend went from a Stripe webhook + Slack alert + spreadsheet to one POST. We process $12k/mo through Rev now and stopped touching reconciliations on Fridays.”
— Founding team, Northstar Labs Read the full case study →
Outcome-based billing

How Rev prices an agent execution

Every meter event is billed from three independent components. You configure the rates, Rev handles the math, tracking, and Stripe settlement.

Component 1

Base fee

Fixed overhead per agent execution. Covers orchestration, queueing, and zero-cost retries.

Component 2

Token rate

Variable compute cost derived from tokens_input and tokens_output on each meter_events row.

Component 3

Outcome bonus

Charged only when outcome === 'success'. Failed or partial runs pay nothing beyond the base and tokens.

price = base_rate + (tokens × token_rate) + outcome_bonus

Rev vs per-call pricing

Pricing alignment Charges per successful outcome — you earn when your customers succeed. Flat rate per call regardless of result.
Customer cost predictability Variable based on success rate; failures cost nothing. Uniform across every call — spikes with retries.
Setup complexity Built-in pricing engine and deferred outcomes — one POST per meter event. Hand-rolled metering, idempotency keys, and outcome reconciliation.
See how Rev's pricing works →
Free
For trying out Rev
$0 /mo
  • 1,000 metering events/month
  • 1 API key
  • Basic analytics
  • Webhook support
  • Community support
Start free
Growth
For growing AI agent products
$199 /mo
  • 100,000 metering events/month
  • 5 API keys
  • Outcome guarantees
  • Advanced analytics
  • Priority support
Get started
Scale
For serious production workloads
$499 /mo
  • 1,000,000 metering events/month
  • Unlimited API keys
  • Outcome guarantees
  • Custom SLA
  • Dedicated support
Get started
Common objections

Top 5 objections to outcome-based billing, answered

How do you prevent gaming of "success" outcomes?
Outcomes are resolved via your own /v1/outcomes call tied to a server-side event your customer actually acknowledged, not a self-report from the agent. Idempotency keys on outcome_id prevent double-counting, and any outcome you resolve as success is stored immutably with the meter event so retrospective edits can't rewrite history.
What if outcomes are hard to define for my agent?
Outcomes don't have to be binary. You can leave outcome_bonus at 0 and bill purely on base + token rate, define outcome as a custom enum (booked, replied, qualified, rejected, expired), or set a generous expiry window so unresolved runs auto-fall to expired and only the cheap components bill.
How is this different from usage-based pricing?
Usage-based pricing bills every request the same flat rate regardless of value. Rev's three-component stack charges a small base + token fee for every run, then an outcome bonus that only posts when the run resolves as success — so you earn on customer wins, not on call volume, and refunds are automatic on failure / expired.
What happens if my agent's confidence is low on a run?
Confidence is your product signal, not Rev's. If your model returns low confidence you decide the outcome state yourself: resolve as failure to skip the bonus, leave it pending until a human reviews it within your expiry window, or treat it as success and tune the bonus rate accordingly. Rev's pricing engine only knows the outcome you POST, never the model's score.
Can I switch tiers mid-cycle?
Yes. Upgrades take effect immediately and are pro-rated by Stripe; downgrades take effect at the next billing cycle. Pricing tier changes never retroactively rewrite historical price_charged values stored on meter_events, so past bills stay auditable.

Not sure which plan fits?

Get a free API key and explore Rev's pricing engine live — no credit card required.

Get Your Free API Key → Try the Playground →

Customer case studies

See all →
Case study
How Northstar Labs cut their billing backend to one POST
Three SaaS products. One Rev integration. $12k/mo billed.
A two-person AI tools team replaced a Stripe webhook + Slack alert + spreadsheet with one POST per agent run. They meter tokens, charge per resolution, and ship new products without touching billing infra.
Read the case study →

Frequently asked questions

Is Rev usage-based pricing?
Rev is outcome-based, not raw usage-based. Every plan includes a monthly base (or event cap) plus a free dev tier; usage above the cap rolls into the next tier rather than billing per call.
How does outcome-based billing compare to per-call pricing?
Per-call charges the same flat rate for every run regardless of result. Rev charges a small base + token fee for every run, then a separate outcome bonus only when the run resolves as success. You earn on wins, not on attempts.
What's included in the Rev free tier?
The free tier ships with 1,000 metering events per month, full access to the metering API and outcome engine, dashboard analytics, and webhook delivery — no credit card required.
Are there per-seat or per-API-key limits on paid plans?
No per-seat fees. Paid plans are capped by monthly metering volume: Starter 10k, Growth 100k, Scale 1M. Multiple API keys are allowed per workspace.
What happens if I exceed my monthly event cap?
Events above the cap continue to be accepted. Over-cap events are tagged for billing in the next period and surfaced in the dashboard so you can either upgrade your tier or throttle at the client.
Can I change plans mid-month?
Yes — upgrades take effect immediately and are pro-rated by Stripe. Downgrades take effect at the next billing cycle.