Free flight API: which of the three kinds you actually need

You typed the words into a search bar because you have a weekend, an idea and no budget. Maybe it is a fare-alert bot, a trip planner for a niche community, or a booking site for the route your family flies every summer. The results look generous: one provider after another with a free tier, a sign-up form and a key in your inbox within minutes. Then, a few days into the build, you find that what you integrated cannot do the one thing your product promised.
That happens because a free flight API is not one category. The phrase returns three different kinds of product, and only one of them lets you sell a ticket. Knowing which one you are looking at before you write the first request is what saves the week.
Three things answer to "free flight API"
Take the marketing away and nearly every result falls into one of these:
| Kind | What a call returns | Free access, typically | Can you sell a ticket with it? | Who carries the money |
|---|---|---|---|---|
| Flight status and schedule data | Departures, arrivals, delays, gates, timetables, sometimes past positions | Common: a capped free tier or a trial key | No. There is no fare and no seat in the response | Nobody. It is information |
| Search-redirect (affiliate) | Prices from a metasearch index and a deep link to someone else's checkout | Common: free to join, paid by commission when a click books | No. The traveller books and pays on the partner's site | The partner site; you earn a commission if the click converts |
| Bookable inventory | Live offers you can search, price, hold, order and issue as an e-ticket | Rarely more than a sandbox; production comes with a contract | Yes. That is the whole point | Whoever issues the ticket, under their own accreditation |
The last column is the real distinction. The first two kinds can be free because nobody's money is at risk when you call them. The third cannot, and the reason has nothing to do with how generous the provider feels.
Status data is useful, and it is not a booking product
A flight data API answers questions about flights that exist: is the evening Dubai departure on time, which gate, what does the timetable on this route look like next month. It is the right tool for a delay notifier, an airport screen, a corporate travel dashboard or a claims product. It is the wrong tool for anything with a price on it. Developers sometimes build a convincing search page on schedule data, add a "book" button, and only then ask where the fare comes from. It does not come from there.
Affiliate search is a business, just not one you control
The search-redirect kind is closer to what most people picture. You send an origin, a destination and dates; you get back prices and a link. When the traveller clicks, they leave your site, book on the partner's checkout, and if the booking completes you earn a commission some weeks later.
That can be a real business, and plenty of content sites run on it. Be clear about what you are giving up. The price you display usually comes from a cached index, so it can differ from what the traveller sees after the click. The customer's name, email and booking reference belong to the partner, not to you. You cannot add a markup, bundle a hotel, offer your own payment method or answer the phone when the flight is cancelled, because there is nothing on your side to change. You are sending traffic, and traffic is the product.
If what you want is a brand that sells travel, with its own customers and its own margin, this kind will feel fine for a month and then become the ceiling.
Why bookable inventory is never free at the ticket
Searching and pricing are read operations. Issuing a ticket is a financial act. Somebody has to hold an IATA ticketing plate (or an airline's direct agreement), issue the e-ticket under it, and settle the money through BSP or the airline's own channel. Whoever that is now owes the airline for the ticket, whether or not your customer's payment holds.
That is where fraud lands. A booking paid with a stolen card is a ticket the issuer has already paid for, and a chargeback weeks later takes the money back from whoever accepted the payment. Fare rules broken through careless automation come back as ADMs (agency debit memos) addressed to the plate holder. So the plate holder wants to know who is sending bookings, from where, with what controls, before your code is allowed to press the button. That review is the contract, and it is why "free" stops at the sandbox.
None of this means a bookable API is out of reach for a small team. It means the cost shows up somewhere: a per-booking fee, a margin built into the net fare you are given, a deposit or a credit limit you work within. Ask which one, early. A provider who cannot tell you how money moves after a ticket is issued has not finished thinking about it either.
What a sandbox should let you do before any contract
A good sandbox is the free part of a bookable API, and it should be enough to tell you whether the integration is worth a commercial conversation. Before you sign anything, check that you can:
- Search real routes. Not one canned response. Try a short-haul route full of low-cost carriers and a long-haul one, and see whether the content differs the way you would expect.
- Price and re-price. Take an offer from search, price it, wait, price it again. Find out how long an offer stays valid and what the response looks like when the price has moved.
- Hold without issuing. Create a booking that reserves seats but does not issue, read back its time limit, then cancel it. If the sandbox only lets you search, you are evaluating the easy half.
- Break things on purpose. Send a name with characters the airline will reject, an infant without an adult, a date in the past. Errors you can read are worth more than any feature list.
- See the post-booking calls. Cancellation, void and refund either exist in the API or go through a support desk. You want to know which before your first live customer asks.
Some of this a sandbox will never show you: real stock running out, a schedule change from the airline, settlement. We wrote a fuller sandbox testing plan covering what to fake and what never to fake, and a walk through the whole booking flow for when you move past it.
Where this platform sits
Plainly: this platform exposes flight and hotel APIs, and they are the bookable kind. Access starts as a conversation rather than a self-serve key. If you need a key tonight to finish a hackathon, it is not us, and it is better you know that from this paragraph than from a sign-up form.
The conversation covers the questions above: what you are building, whether you need ticketing or only search and price, which markets you sell into, how bookings will be paid for. Those answers decide what content you see and how settlement works. If that is where you are, the travel APIs page is where it starts.
What picking the wrong kind costs you
The expensive mistake is not paying for an API. It is building a product on the wrong kind and finding out in front of someone who matters.
It goes like this. The demo runs on an affiliate feed or on schedule data because it was free and fast. The search page looks finished. An investor, a partner agency or your first real customer asks to book, and the only honest answer is that the button leaves the site, or does nothing. Then the rebuild starts, and it is bigger than a provider swap: affiliate results carry no offer id to price, no hold, no passenger model, so the code behind your search page has to change along with the API it calls.
Teams that moved from one bookable provider to another know a gentler version of the same lesson; we covered it in replacing a self-serve flight API. The cheaper path is to decide on the last column of the table first. If your product has to take money for a ticket, start the bookable conversation in week one and build the demo on its sandbox, even if the free options would have got you a prettier screen sooner.