Google Flights Scraper

Scrape Google Flights for live fares, airlines, times, stops, duration, aircraft and CO2 emissions by route and date — keyless, no API key, no browser.

What does Google Flights Scraper do?

This actor reverse-engineers the search payload that Google Flights inlines into its own public results page. It builds the same base64url Protobuf tfs parameter the website uses, fetches the results page with a Chrome TLS fingerprint, and decodes the embedded ds:1 data blob into structured itinerary records. It covers one-way and round-trip searches across any cabin class, passenger mix and stop ceiling, and returns every itinerary Google inlines for that route — typically 15 to 150 options on a busy route.

Output fields

One record per itinerary

Field	Type	Description
origin	string	Origin IATA code from your input
destination	string	Destination IATA code from your input
departure_date	string	Departure date (YYYY-MM-DD)
return_date	string	Return date, or null for one-way
trip_type	string	one-way or round-trip
cabin	string	Cabin class priced
price	number	Total fare in the requested currency, or null when Google has not pre-computed one
currency	string	ISO currency code
airlines	array	Unique marketing carrier IATA codes
airline_name	string	Primary operating airline display name
flight_numbers	array	Carrier + number per leg, e.g. ["AA2413"]
stops	integer	Number of stops (0 = non-stop)
duration_minutes	integer	Total journey duration in minutes
departure_time	string	First leg departure, ISO 8601 local time
arrival_time	string	Final leg arrival, ISO 8601 local time
legs	array	Per-leg detail (see below)
layovers	array	Per-stop airport, city and wait time, or null
co2_emissions_g	integer	Estimated CO2 for this itinerary, grams
co2_emissions_typical_g	integer	Typical CO2 for the route, grams
co2_emissions_delta_pct	integer	Percentage difference from typical
emissions_tag	string	lower, typical or higher
is_best	boolean	Whether Google flagged this as a best result
result_group	string	best or other
booking_token	string	Opaque token Google uses to deep-link this itinerary
source_search	string	Which input search produced this row
scraped_at	string	ISO timestamp

Leg object

Field	Type	Description
airline	string	Marketing carrier IATA code
flight_number	string	Flight number
operating_airline	string	Operating carrier, when different
departure_airport	string	Origin IATA
arrival_airport	string	Destination IATA
departure_airport_name	string	Airport display name
arrival_airport_name	string	Airport display name
departure_time	string	ISO 8601 local time
arrival_time	string	ISO 8601 local time
duration_minutes	integer	Leg duration
aircraft	string	Aircraft type, when Google publishes it
legroom	string	Seat pitch, e.g. 31 in
cabin	string	Cabin for this leg
overnight	boolean	Whether the leg arrives the next day
co2_emissions_g	integer	Leg-level CO2 estimate

Who is it for?

• Fare-monitoring teams tracking a fixed route set daily and alerting on price drops
• Travel agencies benchmarking negotiated fares against the public Google Flights price
• Corporate travel teams enforcing booking-window policy across a route catalogue
• Fare-comparison and metasearch products needing a cheap structured price feed
• Airline and airport analysts studying pricing volatility, stop patterns and capacity
• Data teams building route-level price history for forecasting or emissions reporting

Pricing

Pay per result: $0.15 per 1,000 flight itineraries.

You are charged only for itineraries actually written to the dataset. A run that returns nothing is not billed.

How to use it

1. Open the actor in Apify Console.
2. Set your route: either fill in Origin, Destination and Departure date, or use the Search specs list to run several routes in one run.
3. For a round trip, fill in Return date. Leave it empty for one-way.
4. Choose the cabin class, passenger mix and maximum stops.
5. Set currency, country and language to match the market you want to price.
6. Optionally cap results per search to control cost.
7. Leave the proxy at the default and click Start. Export as JSON, CSV or Excel.

Search spec format: JFK>LAX>2026-11-15 for one-way, JFK>LAX>2026-11-15>2026-11-22 for round-trip.

Output example

{
  "origin": "JFK",
  "destination": "LAX",
  "departure_date": "2026-11-15",
  "return_date": null,
  "trip_type": "one-way",
  "cabin": "economy",
  "price": 189,
  "currency": "USD",
  "airlines": ["AA"],
  "airline_name": "American Airlines",
  "flight_numbers": ["AA2413"],
  "stops": 0,
  "duration_minutes": 380,
  "departure_time": "2026-11-15T08:15:00",
  "arrival_time": "2026-11-15T11:35:00",
  "legs": [
    {
      "airline": "AA",
      "flight_number": "2413",
      "operating_airline": null,
      "departure_airport": "JFK",
      "arrival_airport": "LAX",
      "departure_airport_name": "John F. Kennedy International Airport",
      "arrival_airport_name": "Los Angeles International Airport",
      "departure_time": "2026-11-15T08:15:00",
      "arrival_time": "2026-11-15T11:35:00",
      "duration_minutes": 380,
      "aircraft": "Airbus A321neo",
      "legroom": "31 in",
      "cabin": "economy",
      "overnight": false,
      "co2_emissions_g": 312000
    }
  ],
  "layovers": null,
  "co2_emissions_g": 312000,
  "co2_emissions_typical_g": 340000,
  "co2_emissions_delta_pct": -8,
  "emissions_tag": "lower",
  "is_best": true,
  "result_group": "best",
  "booking_token": "CjRIQ...",
  "source_search": "JFK->LAX 2026-11-15",
  "scraped_at": "2026-09-28T12:00:00.000Z"
}

Technical details

• Stack: Node.js 24, apify SDK, impit HTTP client with Chrome TLS impersonation, tough-cookie for the consent cookie.
• Transport: the public Google Flights results page. Google's GetShoppingResults RPC endpoint has been gated since August 2026 by an x-goog-batchexecute-bgr header that only the page's own JavaScript can produce, so this actor reads the same rows from the page's inline ds:1 data blob instead.
• Encoding: the tfs URL parameter is a hand-rolled Protobuf message carrying trip type, per-direction date, origin/destination airports, cabin, stop ceiling and passenger codes. The encoder is 40 lines with no protobuf runtime dependency.
• Consent: a pre-accepted SOCS cookie is sent on every request, which skips Google's EU consent interstitial.
• Retries: a 200 response that carries no ds:1 blob is a known transient page variant and is retried up to three times with backoff. HTTP 429 and 503 are also retried; other HTTP errors fail fast.
• Proxy: defaults to the Apify datacenter pool, which is sufficient when the request carries a Chrome TLS fingerprint. Residential proxy is available by changing the proxy input.

Known limits

• Multi-city itineraries are not supported. Google loads multi-city results client-side through the gated RPC, so the page carries no rows to read. Search each leg separately.
• Page 1 only. Google inlines every row it is willing to show for a route on one page. There is no deeper pagination to fetch, so a search returns what Google shows a browser on first load — typically 15 to 150 itineraries.
• Premium-cabin and child fares are often unpriced. Google prices parties with children or infants client-side and frequently omits an aggregate fare for premium-cabin round trips. Those rows come through with price: null and a populated booking_token.
• Prices are snapshots. Google's fares move on inventory buckets and can change within minutes. Treat each row as a point-in-time observation.
• Low-traffic airports may return fewer rows. The actor sets the on-demand pricing flag Google uses for regional airports, but very thin routes can still come back with little or nothing.
• Google can change the page shape. The ds:1 blob key and the row index layout are reverse-engineered. If Google reshapes the payload, the actor fails loudly with the marker line in the log rather than silently returning wrong data.

FAQ

Do I need a Google account or API key? No. The actor reads the same public results page a logged-out browser sees. No login, no token, no developer account.

Why is there no official Google Flights API? Google retired the public QPX Express API in April 2018 and now offers airfare data only through enterprise contracts. Scraping the consumer site is the only programmatic route for everyone else.

Why does my run return zero results? Open the run log and find the ds:1 marker line. If it shows a consent interstitial, a sorry page or a captcha, switch the proxy input to residential and re-run. If it shows AF_initDataCallback=true but key 'ds:1'=false, Google has reshaped the page and the actor needs an update.

Can I get business class prices? Yes — set the cabin class to business. Note that Google often omits an aggregate price for premium-cabin round trips, so you may see price: null on those rows even though the itinerary is returned.

How many flights do I get per search? One page load, which is everything Google inlines: typically 15 to 30 on a thin route, 100 or more on a busy one. There is no way to fetch more for the same search without changing the search.

Does it support one-way and round-trip? Yes, both. Round-trip prices are the combined return fare. Multi-city is not supported — see Known limits.

Support

Open an issue on the Actor's page for bugs or feature requests.

Changelog

0.1 — 2026-09-28

• Initial build. Keyless extraction of Google Flights results from the public search page.
• Hand-rolled Protobuf tfs encoder (trip type, dates, airports, cabin, stops, passengers) with no protobuf runtime dependency.
• AF_initDataCallback ds:1 payload extraction and full itinerary decoding: price, currency, per-leg airline, flight number, airports, times, aircraft, legroom, cabin, layovers, CO2 and emissions tag.
• SOCS consent cookie sent on every request to skip the EU interstitial.
• Retry on the transient no-ds:1 page variant, on HTTP 429 and on HTTP 503.
• One-way and round-trip support; multi-city explicitly rejected with a clear message.
• Single-route fields plus a searches list for multi-route runs in one execution.
• Defaults to Apify datacenter proxy; residential available via the proxy input.

Changelog

0.1 — 2026-09-28

• Initial build. Keyless extraction of Google Flights results from the public search page.
• Hand-rolled Protobuf tfs encoder (trip type, dates, airports, cabin, stops, passengers) with no protobuf runtime dependency.
• AF_initDataCallback ds:1 payload extraction and full itinerary decoding: price, currency, per-leg airline, flight number, airports, times, aircraft, legroom, cabin, layovers, CO2 and emissions tag.
• SOCS consent cookie sent on every request to skip the EU interstitial.
• Retry on the transient no-ds:1 page variant, on HTTP 429 and on HTTP 503.
• One-way and round-trip support; multi-city explicitly rejected with a clear message.
• Single-route fields plus a searches list for multi-route runs in one execution.
• Defaults to Apify datacenter proxy; residential available via the proxy input.

Changelog

0.1 — 2026-09-28

• Initial build. Keyless extraction of Google Flights results from the public search page.
• Hand-rolled Protobuf tfs encoder (trip type, dates, airports, cabin, stops, passengers) with no protobuf runtime dependency.
• AF_initDataCallback ds:1 payload extraction and full itinerary decoding: price, currency, per-leg airline, flight number, airports, times, aircraft, legroom, cabin, layovers, CO2 and emissions tag.
• SOCS consent cookie sent on every request to skip the EU interstitial.
• Retry on the transient no-ds:1 page variant, on HTTP 429 and on HTTP 503.
• One-way and round-trip support; multi-city explicitly rejected with a clear message.
• Single-route fields plus a searches list for multi-route runs in one execution.
• Defaults to Apify datacenter proxy; residential available via the proxy input.