FlightByte
Benchmarks · Independent tests

Tested by others, caveats included.

Tests of FlightByte that other people ran and published. We list each one with its method and its limits, including the results where another provider came out ahead.

These tests were not run by FlightByte.

Every benchmark on this page was designed, run and published by someone outside FlightByte. We did not commission or pay for any of them, and we have no business relationship with their authors. The figures and charts are theirs, as published, with a link to the original write-up.

[ INDEPENDENT · NOT RUN BY FLIGHTBYTE ]

20 flights across six continents

By Nebi Berke İçöz on Medium, published Nov 12, 2026. Test day: Nov 11, 2026. Designed and run by the author; FlightByte had no part in it.

Read the full write-up
Touchdowns reported first by FlightByte
19 of 20
Confirmed gate arrivals reported first
17 of 20
Median lead over AirLabs on touchdown, same flight
6.3 min
Median ETA error 30 minutes before arrival
2 min

How it was run

FlightByte
Flight subscriptions created before the test day; changes arrived as webhooks.
AirLabs
Standard Flight Information REST API, polled. Its beta Flight Alert webhooks were not tested.
AeroDataBox
Flight Status (single day) on API.Market, polled.
Polling
From 180 minutes before scheduled arrival: every 15 minutes, then every 3 minutes in the final hour.
Flights
20 arrivals at 20 different airports in Asia-Pacific, the Middle East, Europe, Africa, South America and North America.

Results

Results of 20 flights across six continents
MetricFlightByteAirLabsAeroDataBox
Touchdown observed20/2020/2016/20
First to report touchdown19/200/201/20
Median time after touchdown3.2 min8.9 min7.3 min*
Touchdown, 90th percentile4.2 min11.8 min15.0 min*
Confirmed gate arrival observed20/2017/206/20
First to report confirmed gate arrival17/200/203/20
Median time after gate arrival5.3 min12.9 min5.7 min*
Gate arrival, 90th percentile8.4 min16.1 min11.9 min*

* AeroDataBox medians use only the flights where it reported a confirmed event. Three of its “Arrived” answers carried a gate time that was still in the future and are not counted as arrivals. Times are from the event to the moment the test harness first saw it.

The charts

Bar chart of minutes after touchdown until each integration saw it: FlightByte median 3.2, 90th percentile 4.2; AirLabs 8.9 and 11.8; AeroDataBox 7.3 and 15.0.
Touchdown: minutes until the integration saw the event. Median bar, 90th percentile marker. Chart: Nebi Berke İçöz, Medium
Bar chart of minutes after gate arrival until each integration saw it: FlightByte median 5.3, 90th percentile 8.4; AirLabs 12.9 and 16.1; AeroDataBox 5.7 and 11.9 over six flights.
Gate arrival: the same measure. AeroDataBox over its six confirmed arrivals. Chart: Nebi Berke İçöz, Medium
Dot plot of minutes AirLabs saw each event after FlightByte, per flight. Every touchdown and gate-arrival gap is positive, between about 3 and 9 minutes.
The same flight, the same event: how many minutes after FlightByte the AirLabs integration saw it. Every point is to the right of zero. Chart: Nebi Berke İçöz, Medium
Bar chart of confirmed events seen: touchdown 20, 20 and 16 of 20; confirmed gate arrival 20, 17 and 6 of 20, for FlightByte, AirLabs and AeroDataBox.
Confirmed events seen within the test window. A predicted gate time does not count as an arrival. Chart: Nebi Berke İçöz, Medium
Bar chart of median absolute ETA error at 90, 60 and 30 minutes before scheduled arrival: FlightByte 5.0, 3.5 and 2.0 minutes; AirLabs 6.0, 4.0 and 4.0; AeroDataBox 7.5, 7.0 and 7.0.
Median ETA error at three checkpoints, measured against FlightByte’s first gate time. Chart: Nebi Berke İçöz, Medium
Timeline for AA764: at 20:43 AeroDataBox marked the flight Arrived with a gate time of 20:49, still in the future; FlightByte's gate time was 21:01 and its webhook arrived at 21:03.
AA764: “Arrived” with a gate time still in the future. Times in Türkiye local time. Chart: Nebi Berke İçöz, Medium
Bar chart of median minutes after touchdown by region, FlightByte against AirLabs: Asia-Pacific 3.2 and 8.9, Middle East 3.7 and 10.1, Europe 2.0 and 8.8, Africa and South America 4.1 and 7.2, North America 3.1 and 10.2.
Touchdown by region. Some groups hold two flights, so read them as a description, not a regional ranking. Chart: Nebi Berke İçöz, Medium
Bar chart of the share of flights with a filled field: actual runway time 100, 0 and 65 percent; actual gate time 95, 85 and 45; gate number 55, 55 and 25; baggage carousel 60, 60 and 50.
Fields that were filled in. A filled field is not proof the value was right or on time. Chart: Nebi Berke İçöz, Medium

Where others led

  • AeroDataBox reported touchdown first on AA764, Houston to Chicago O’Hare.
  • AeroDataBox reported gate arrival first on SV596 into Dubai, QR831 into Doha and AC707 into Toronto: three of the six gate arrivals it confirmed.

What this test doesn’t show

  • AirLabs was polled, every 3 minutes in the final hour. Its beta Flight Alert webhooks, which need a paid plan, were not part of the test, so this is not a webhook-to-webhook comparison.
  • There was no independent ground truth: touchdown and gate times were taken from FlightByte’s own fields. The paired, same-flight gaps do not depend on that reference.
  • One day and 20 chosen flights, not a random sample. It is not a ranking of flight data providers or a statement about every airport.
  • Only AeroDataBox’s single-day Flight Status product was tested, not its other endpoints or its alerts.
[ YOUR TEST ]

Ran your own test?

If you benchmarked FlightByte against other flight APIs and published how you did it, send us the link. We list it whatever the result.

What we look for

  • The dates and the flights are stated.
  • How each API was called is stated: webhook or polling, and how often.
  • The timings are published, so others can check them.
  • Any tie to FlightByte is disclosed, free credits included.
Get started

Track your first flight tonight.

An API key, the docs, and two requests stand between you and your first webhook.

Get API keySee the plans

A free key in a minute, no card. Paid plans are paid by card in the dashboard.