Priced like infrastructure, not a taxi meter — pay only for the requests you run, with no surprise overages, ever.
Pricing FAQ
What is a request unit? A request unit is any request used to trigger your config — whether that be an API, workflow, CLI etc. It includes 1000ms of execution time. Your requests can run as many actions as needed, and as long as they need, either bound by your own limits if self hosted or ours if hosted on our infrastructure. Why 1000ms? A good API response time would be beneath 300ms, and anything past ~850ms would be considered poor performing. This allows us to cater for the most typical understanding of a “request”. Usecases may differ but this allows us to simplify pricing and make usage easy to understand.
Is this a monthly subscription? No. Request units are pay-as-you-go — buy what you need and draw them down as you go. Your free 3,000 managed + 1,000 self-hosted units refresh each month; anything you pay for is yours to keep. Platform isn't a subscription or a fee either — it unlocks while your account holds at least 100,000 paid units (about $199/mo of usage). Any top-ups count toward it, the units are still yours to spend, and they roll over. You're always paying for units, never for status.
Need a bigger or custom package? Just ask — we'll sort it out. Custom sizes, volume rates, different managed/self-hosted splits, invoicing or bank transfer: contact us and we'll set up a package that appears directly in your account, ready to purchase. If you're running serious volume (hundreds of thousands of units and up), that's exactly the conversation we want to have.
Do request units roll over? Yes — unused paid units roll over so you never lose what you've bought, for up to 12 months from purchase. Your free monthly allowance refreshes at the start of each month and doesn't stack — and when a request runs, free units are always drawn down before paid ones.
What is the difference between AP-Hosted & Self-Hosted request units? If you are running your config on our backend, they are deducted from your hosted/managed units balance. If you are self hosting your configs, they are deducted from your self hosted/unmanaged units balance. You are allowed to use both offerings / units simultaneously to suit your needs.
What is a config? An Air Pipe config is a simple YAML file — you can define interfaces to trigger such as HTTP routes, or have it operate as a CLI based tool etc. These interfaces can have many actions configured to fetch and process data.
What is an action? An action could be: fetch data from this HTTP endpoint, run this command, make this database query etc. You can use any number or any combination of them in your config.
How do you process payments? We use AirWallex payment links, and for other customer cases we can accept bank transfers in particular currencies on a case by case basis. https://www.airwallex.com/
Scenario: My config or API request ran for 900ms That's one request unit. Anything up to a full second is a single unit — we bill in whole seconds and round in your favor. No fractions, no multipliers, no per-step math.
Scenario: My config or API request ran for 4900ms Five request units — one per second. It doesn't matter how many actions, connections, or AI calls ran inside it; you're billed by time, not complexity.
Scenario: My config needs to make 100 actions, and connect to many DBs and APIs We don't care how many actions or connections you make. You're billed by request units: under a second is one unit, five seconds is five. 100 actions across ten databases in 800ms? One unit.
What happens if I run out of units? Your workflows don't halt without warning. We alert you as you approach your balance, and you can top up or roll into usage — no silent shut-offs, no lost runs, no surprise overages.