Blog / API guides

Kiwi.com API: Search Fares and Cheapest Dates as JSON

7 Sept 2026Perceptlayer6 min read

Kiwi.com's own API is now for invited business partners, so most developers can't get a key for it. To read Kiwi.com fares as JSON, call GET /v1/kiwi/flights-search with origin, destination and departure_date; each successful call costs 1 credit and returns itineraries with price, carriers, stops, duration, baggage and a booking link. For a date window, GET /v1/kiwi/calendar-cheapest-dates takes a start_date and end_date in one call.

Can you still get a Kiwi.com API key?

Not unless Kiwi.com invites you. Kiwi.com used to give public access to its Tequila platform, which included the API, a white-label product and an affiliate programme. It has since narrowed that to key business partnerships, and new Tequila partnerships are by invitation only; existing partners keep their tools and documentation (Kiwi.com, 2024).

If you're a travel agency that wants to sell Kiwi.com itineraries and take bookings, the partner route is the right one: the Tequila site asks prospective affiliates to get in touch with Kiwi.com by email (2026). If you want to read fares for research, alerts or an agent, the Kiwi.com API returns what the site shows, as JSON, with the same key as every other API in the catalogue. It doesn't book anything: each itinerary carries a booking_url that opens the booking page on Kiwi.com.

What can the Kiwi.com API return?

Seven GET endpoints cover search, a single itinerary, date flexibility and destination discovery. Each costs 1 credit per successful call. The two that matter most are flights-search for one route on one date and calendar-cheapest-dates for one route across a window. locations turns a place name into the codes the other endpoints take.

explore-price-map also takes a bounding box (south_west_lat, south_west_lng, north_east_lat, north_east_lng) and a limit, which is how you ask "where can I fly from London in December" for one region of the map.

How do you search Kiwi.com fares?

Send origin, destination and departure_date to GET /v1/kiwi/flights-search. Codes can be airports or cities: the recorded request below uses LON for London and BCN for Barcelona. Add return_date for a return trip. If you don't know a code, call /v1/kiwi/locations?query=London first. This is the request behind the recorded response in the next section:

curl "https://api.perceptlayer.com/v1/kiwi/flights-search?origin=LON&destination=BCN&departure_date=2026-12-04" \
  -H "X-API-Key: $API_KEY"

You need a key in the X-API-Key header; get a free API key, and the authentication docs cover the rest.

Which parameters does flights-search take?

Three are required and fourteen optional. The table lists them with the types and allowed values from the endpoint definition. Nothing else is documented about defaults, although the recorded response echoes the values it used, such as cabin economy, currency EUR and sort quality.

To see only direct flights on one airline, combine max_stops=0 with carriers[]=FR.

What does a flights-search response look like?

The response echoes the search, gives total_available, and lists itineraries. The recorded London to Barcelona search on 2026-12-04 found 23. Here is the first itinerary, trimmed with …:

{
  "success": true,
  "data": {
    "trip_type": "one_way",
    "origin": "LON",
    "destination": "BCN",
    "departure_date": "2026-12-04",
    "cabin": "economy",
    "currency": "EUR",
    "sort": "quality",
    "total_available": 23,
    "itineraries": [
      {
        "id": "ItineraryOneWay:eyJwcm92aWRlcnMiOiJLSVdJLUJBU0lDOktJV0kiLCJyb3V0ZV9kYXRhIjoiRlI6NzgwNzpMVE46MTc5NjM5ODUwMDpCQ046MTc5NjQwNjMwMDplY29ub215OkZhbHNlOjo6IiwicHJpY2Ui…",
        "price": { "amount": 29, "currency": "EUR", "formatted": "29 €" },
        "duration": 7800,
        "stops": 0,
        "carriers": [{ "code": "FR", "name": "Ryanair", "icao": "RYR" }],
        "departure": {
          "airport_code": "LTN",
          "city": "London",
          "local_time": "2026-12-04T15:35:00"
        },
        "arrival": {
          "airport_code": "BCN",
          "city": "Barcelona",
          "local_time": "2026-12-04T18:45:00"
        },
        "segments": [{ "flight_number": "FR7807", "cabin_class": "ECONOMY", "duration": 7800, … }],
        "booking_url": "https://www.kiwi.com/en/booking/?direct=true&locale=en&currency=eur&passengers=1-0-0&token=…",
        …
      },
      …
    ]
  },
  …
}

The second itinerary is FR8216 from Stansted (STN) at 32 EUR, which is how you know LON searched every London airport. Times are local to each airport, so compute flight time from duration rather than from the two timestamps. Each itinerary's id is what flights-detail takes as itinerary_id, up to 2,000 characters. For what the baggage block says about the real price of a cheap fare, see what fare tracking can and cannot show.

How do you find the cheapest date to fly?

Use GET /v1/kiwi/calendar-cheapest-dates with a route and a window. It takes origin, destination, start_date and end_date, plus optional nights_min and nights_max for a return stay, and returns the cheapest dates in that window in one call. We don't have a recorded response for it, so run it in the playground on the Kiwi.com API page to see its fields and how wide a window it accepts.

curl "https://api.perceptlayer.com/v1/kiwi/calendar-cheapest-dates?origin=LON&destination=BCN&start_date=2026-12-01&end_date=2026-12-30&currency=EUR" \
  -H "X-API-Key: $API_KEY"

When you need full itineraries for every date, not only the cheapest, loop over flights-search instead. This script scans 30 departure dates and keeps the lowest price.amount for each, using only fields from the recorded response:

import datetime as dt
import os
import requests

URL = "https://api.perceptlayer.com/v1/kiwi/flights-search"
HEADERS = {"X-API-Key": os.environ["API_KEY"]}
start = dt.date(2026, 12, 1)

for i in range(30):
    day = (start + dt.timedelta(days=i)).isoformat()
    r = requests.get(URL, params={"origin": "LON", "destination": "BCN",
                                  "departure_date": day, "currency": "EUR"},
                     headers=HEADERS, timeout=60)
    if r.status_code != 200:
        print(day, "error", r.json()["error"]["code"])
        continue
    its = r.json()["data"]["itineraries"]
    if not its:
        print(day, "no itineraries")
        continue
    best = min(its, key=lambda x: x["price"]["amount"])
    print(day, best["price"]["amount"], best["carriers"][0]["code"],
          best["departure"]["airport_code"], best["duration"] // 60, "min")

The script reads only fields from the recording, but we haven't run it live for this article. Each line prints the date, the lowest fare, the airline, the departure airport and the flight time in minutes. A day that fails prints its error code and costs nothing.

What does a 30-day date scan cost?

Every Kiwi.com endpoint costs 1 credit per successful call, so the cost depends on how many calls the scan takes. With flights-search, one call per departure date makes a 30-day scan 30 × 1 = 30 credits per route. With calendar-cheapest-dates, the same window is one call, 1 credit, if the endpoint accepts a 30-day range.

The last row is the usual compromise: let the calendar find the dates, then pull full itineraries only for the ones worth showing. Watching 10 routes that way daily is 120 × 10 = 1,200 credits a month, which is 1,200 − 1,000 = 200 credits a month beyond the free tier. Failed calls return X-Credits-Used: 0. See the credits docs for the balance and pricing for what packs cost.

To try it, run the London to Barcelona search above in the playground on the Kiwi.com API page, then change the date.

kiwi.comflightsflight pricestravelcheapest dates