Will you be able to work on that flight?
Type the flight number and you get three things: the odds it draws next-gen WiFi, what is on board when it does not, and how far to lean on both. Every figure on this site names its source and its date. No account, nothing stored. The page runs in two halves — the record first, then the companion, a Chrome extension that carries the same odds onto the page where you book.
The live check needs JavaScript. Pick your airline instead. Every page below carries the same ConnectScore, the method behind it, and how much to trust it.
UA2451 · ORD → IAH
Odds this departure draws a Starlink aircraft
Starlink came up on 5% of this flight’s last 10 departures (tracker history, 25 Jul 2026). Plan the trip around the older cabin, and carry something to do offline.
The umbrella: aircraft assignments firm up about two days out, so re-check at T-48h. If you drew a dud, UA1799 on the same route pulled a Starlink tail on 61% of its last 10 departures, and the route optimizer ranks all 4 of them. Rank this route →
Reported · flight history from unitedstarlinktracker.com, 25 JUL 2026 · confidence medium, 10 observations
Where all 18 airlines stand
the sortable leaderboard →3 fleets are finished, 12 are mid-retrofit and 3 have signed for hardware that has not flown, counted 25 Jul 2026. The middle of a rollout is where odds are worth having: before it everyone is on old hardware, and after it the question dies of success.
The seven a US traveller is most likely to board, best next-gen odds first
Hawaiian
64goodNext-gen: 64 · Starlink on 42 of 66 (64%)
Today: Starlink 64%, rest streaming-class or basic · free onboard
ConnectScore 64 · good
United
30long shotNext-gen: 30 · Starlink on 481 of 1,579 with a published system (30%)
Today: Starlink 30%, rest streaming-class or basic · free for loyalty members
ConnectScore 48 · mixed
Alaska
28long shotNext-gen: 28 · Starlink on 99 of 350 (28%)
Today: Starlink 28%, rest streaming-class · free onboard
ConnectScore 55 · mixed
jetBlue
0not yetNext-gen: 0 (Amazon Leo signed, 2027)
Today: streaming-class fleetwide · free onboard
ConnectScore 55 · mixed
American
0not yetNext-gen: 0 (Starlink signed, 2027-Q1)
Today: streaming-class on 90% of the fleet · the rest basic or none · free onboard
ConnectScore 51 · mixed
Delta
0not yetNext-gen: 0 (Amazon Leo signed, 2028)
Today: streaming-class on 86% of the fleet · the rest basic or none · free onboard
ConnectScore 49 · mixed
Southwest
0not yetNext-gen: 0 · Starlink on 1 of 803 (<1%)
Today: Starlink <1%, rest streaming-class or basic · free for loyalty members
ConnectScore 16 · long shot
| # | Airline | ConnectScore | Tier | Next-gen odds | Rollout | Projected · grey, never sorts |
|---|---|---|---|---|---|---|
| 01 | airBalticBT | 100 good | C | 100 | Finished | · |
| 02 | JSXXE | 100 good | C | 100 | Finished | · |
| 03 | ZIPAIRZG | 100 good | C | 100 | Finished | · |
| 04 | Air FranceAF | 80 good | C | 75 | Installing now | · |
| 05 | WestJetWS | 79 good | C | 79 | Installing now | · |
| 06 | HawaiianHA | 64 good | C | 64 | Installing now | · |
| 07 | Qatar AirwaysQR | 61 good | C | 58 | Installing now | · |
| 08 | AlaskaAS | 55 mixed | B | 28 | Installing now | · |
| 09 | jetBlueB6 | 55 mixed | D | 0 | Signed, nothing flying | 25begins 2027, complete 2028SOFT |
| 10 | SASSK | 52 mixed | C | 49 | Installing now | · |
| 11 | AmericanAA | 51 mixed | D | 0 | Signed, nothing flying | 51installs begin 2027-Q1FIRM |
| 12 | DeltaDL | 49 mixed | D | 0 | Signed, nothing flying | 38begins 2028FIRM |
| 13 | UnitedUA | 48 mixed | A | 30 | Installing now | · |
| 14 | Virgin AtlanticVS | 39 long shot | C | 28 | Installing now | · |
| 15 | EmiratesEK | 31 long shot | C | 16 | Installing now | · |
| 16 | Air CanadaAC | 18 long shot | C | 6 | Installing now | · |
| 17 | SouthwestWN | 16 long shot | D | 0 | Installing now | 37300+ by end-2026, installs runningFIRM |
| 18 | British AirwaysBA | 15 long shot | C | 2 | Installing now | · |
Ranked on what is flying today, 25 Jul 2026. Bands: good 60 to 100, mixed 40 to 59, long shot 1 to 39, not yet 0. The tier column is how the row was derived: A resolves your aircraft to a registration, B stops at the aircraft type, C is the airline’s own published fleet count and D is a signed deal with nothing in the air. ConnectScore is an expected value over a whole fleet, not a prediction about one flight: United measured 320, 56 and 15 Mbps on three systems in one livery in one reporting period. Aircraft whose system an airline does not publish are left out of the denominator rather than assumed. Signed-but-unflown deals (AA Starlink 2027, DL/B6 Amazon Leo) score zero until they fly.
Reported Tail verification for United, Alaska and Hawaiian: unitedstarlinktracker.com and alaskastarlinktracker.com, both by @martinamps, 25 JUL 2026. Every other airline from public airline announcements, Jul 2026. The full credit.
The confidence ladder, and the fourth number
the full method →A 27 read off a tail record and a 27 read off a press release are different claims about the world. Four tiers keep them apart, every answer names its tier, and the forward-looking number lives behind a fence of its own.
ATail-verified
The aircraft on your flight is resolved to a registration, and that registration to an install record. A number for one flight, with the sample size attached.
United
BType-derived
The tails are verified the same way and nobody publishes which aircraft is scheduled onto which flight, so there is no history to count. A number for an aircraft type. Not for a departure.
Alaska
CFleet-share
No per-tail verification exists, so the input is what the airline itself said publicly about how many aircraft are equipped. A number for an airline. Nothing narrower.
airBaltic, JSX, ZIPAIR, Air France, WestJet, Hawaiian, Qatar Airways, jetBlue, SAS, American, Delta, Virgin Atlantic, Emirates, Air Canada, Southwest, British Airways
DAnnounced only
A signed deal with nothing in the air. Today’s odds are zero and the projected figure carries the forward view, fenced, in grey. A promise with a year on it. Never a measurement.
jetBlue, American, Delta, Southwest
38begins 2028FIRM
Where a carrier has signed and published an aircraft count, a projected score appears in that grey chip and nowhere else. It never sorts a table. It carries its horizon date and a confidence word everywhere it goes, and when a horizon passes with nothing installed it flips to SLIPPED while keeping the date it missed. That flip is computed from the build date, so it does not wait for anyone to notice. The arithmetic and all five rules →
Three limits hold everywhere. Tail swaps happen up to pushback, which is why the T-48h re-check is in every playbook. This site tracks aircraft, never seats. And nobody has load-tested a full cabin, so every crowding claim, from anyone, is inference — I think that is the largest open question in the subject.
Measured · the load-test gap holds across Jang et al., ACM IMC ’25, Oct 2025 · Ullah et al., arXiv:2508.09839, Aug 2025 · Ookla Speedtest Intelligence, 28 Apr 2026
A badge on every flight in the results
WiFi Odds for Flights is a Chrome extension, free in the Chrome Web Store under that name. On a united.com or app.navan.com results page every flight picks up an odds badge, one click re-sorts the page by odds while prices and times stay put, and a panel ranks the route’s departures the way the answer card above does. No account and no tracking, there or here. The pictures are captures of the shipped build.
- 1Search
You search united.com the way you always do. The extension adds nothing to the page until there are results on it, and it never touches the booking or the payment steps.
- 2Odds
Every result picks up its own badge: how often that flight number has actually drawn a Starlink aircraft. 49% here, 44% on the next one. Once the assigned tail is confirmed equipped the badge shows a ✓ instead of a percentage.
- 3Sort
One click reorders the whole page best-odds-first, prices and times untouched, so you can see what the good connection actually costs you. Same panel on Navan as on united.com.
- 4Guard
The panel names what it cannot know yet: tails firm up around 48 hours out. Watching that window after you book is the Tail-swap Guardian, which is built, in test, and ships in 2.1.
Captured on the real united.com and Navan with the store build. Nothing here is a mockup. The percentages in the badges are the odds the extension actually computed for that route.
Reported · departure histories from unitedstarlinktracker.com, 25 JUL 2026 · badges use the board’s bands and the same numbers this site publishes, so when the record is wrong the badge is wrong the same way
- In the store v1.5.1 covers united.com and app.navan.com. A badge carries a tick once the assigned tail is confirmed equipped, and the percentage until then. Checked 25 Jul 2026.
- Submitted, in review v2.0.0 adds alaskaair.com and Google Flights, each behind a permission you grant in the popup and off until you do, plus an 18-airline odds popup. Not live until the store says so, so installing today gets you none of it.
- Built, unreleased Tail-swap Guardian watches a booked flight for an equipment swap between booking and boarding. In no store build yet; its horizon is on the roadmap, dated and fenced like everything else here.
No accounts, no analytics, no tracking. It keeps your settings on your own machine and phones nothing home. Unofficial, and not affiliated with any airline, Navan, SpaceX/Starlink or the trackers.
Reported Store version and coverage read off the Chrome Web Store listing body, 24 Jul 2026, rather than off the repository manifest, which is already at 2.0.0.
What the record cannot see, you can
Fleet records stop at the cabin door. Whether the login worked, what a video call survived, which badge was wrong: the person in the seat is the only instrument on board, and this site runs no analytics to guess with. Five channels carry observations back, and none of them asks for your email.
- The report form
- At the foot of every airline page, posting to the same intake the extension uses. A flight, a date, what the WiFi did. Plain feedback about the site or the extension rides the same channel.
- The extension itself
- The panel carries a report action, and the design calls for one accuracy prompt after a few real uses. Never on install day, never twice.
- A store review
- Chrome ranks listings partly on reviews, so a review is the one act that puts the extension in front of the next traveller. If it earned one, two sentences there carry more weight than anything I could write about it myself.
- GitHub issues
- Bugs, wrong badges with a screenshot, feature requests. Slower than the form, better for anything with a repro.
- Field reports keep surfacing in airline and Starlink threads. I read them, and a confirmed one enters the record wearing its REPORTED label like everything else.
The intake · POST /api/report · no channel attaches an account · what the server keeps is on the privacy page