The real landing time,
one hour ahead.
THOR computes when each arriving aircraft will actually land, from its position in flight, and recomputes it every minute. Measured on real traffic: a mean error of 3 minutes 15 minutes before landing, 9 minutes one hour before — against landing times actually observed, published airport by airport.
Provider-independent by design. Production data sources are qualified on performance, coverage, freshness, contractual rights and reliability.
The sky moves. Ground operations react too late.
A prettier ETA is not enough. Teams need to know which arrivals are moving materially, when the signal changed, and whether it is strong enough to deserve attention.
Crew & equipment
Teams position too early or too late when arrival expectations move faster than operational planning.
Stands & buses
Late arrival changes cascade into stand reviews, bus coordination and peak handling pressure.
Trucks & docks
Arrival uncertainty affects dock readiness, truck timing and SLA-sensitive night waves.
Too much noise
Operators monitor too many flights and too many changing signals without an explicit attention queue.
Know one hour ahead, to the minute, instead of finding out at the gate.
Today the ground works from the airline’s announced time, which rarely moves until the aircraft is close. THOR starts from where the aircraft is now.
Handlers
Crews and equipment positioned on the real time: no idle wait on an aircraft announced on time that lands 20 minutes later, no scramble on one that arrives early.
Airports
Stands, buses and gates re-planned one hour ahead, on an attention queue of the arrivals that actually move — not on every flight.
Cargo
Trucks and docks called on a landing time that holds to a few minutes, for the night waves where every slot is contractual.
Management
Every prediction and every action taken by the team are logged, time-stamped and exportable: you can prove what THOR changed.
Side by side, on facts
| Airline announced time | Public flight trackers | THOR Flights | |
|---|---|---|---|
| Time source | Flight plan, updated by the airline | Aircraft position | Aircraft position, THOR engine (FIX2 + CAL1, locked by fingerprint) |
| Refresh | Occasional | Minutes | Every minute, each arrival |
| Accuracy measured and published | No | No | Yes — per airport and horizon, against observed landings, sealed weekly |
| Compared on your own flights | — | — | Yes — announced time vs THOR vs actual landing, visible during the free demo |
| Attention queue | No | No | Yes — which arrivals move materially, and since when |
| Action log for the ground team | No | No | Yes — who did what, on which prediction, exportable |
| API and integration | Depends | Not for operations | API, CSV, web space; AODB/FIDS optional |
| Use | Reference | Information | Ground-resource planning (never safety-of-flight) |
Generic categories, not named products. THOR’s figures: live proof page. The announced-time comparison is measured on your own arrivals during the demo and the Proof Sprint.
Real arrivals. Real THOR predictions.
This board is fed by the THOR engine running now on <span id="live-airports">the showcase airports</span>. Flight identifiers are masked, no aircraft position is ever shown, and the board refreshes every 15 minutes.
Measured first. Claimed second.
THOR Flights is sold as an evidence-backed operational product, not as a promise of perfect prediction.
Latest 7-day measurement of the best-covered airport, read live from the proof page (Nice, 10–17 Oct 2026 at publication). One error per flight: the last time THOR announced at least H minutes before the landing actually observed. Nothing removed afterwards.
Current evidence level. Predictive performance (P2) is measured every week on live traffic against observed landing times, and sealed. Operational utility (P3) and economic value (P4) are measured on your station through a Proof Sprint. Historical P2 cohort (95 flights, 0–2 h): THOR 7.98 min vs 14.95 min for the reference, THOR better on 65.3% of flights.
Any airport. Live within 30 minutes.
Search among the world’s airports with scheduled traffic. THOR can follow any of them: request a free 7-day live demo and see your own arrivals.
Data in. THOR decisions out.
THOR Flights is designed around qualified commercial aviation data and the customer’s own operational truth. Supplier data is an input. THOR outputs are the product.
Aircraft state
Position, altitude, speed, track and observation time under appropriate contractual rights.
Normalize & qualify
Freshness gates, source abstraction, active-flight selection and customer operational context.
Arrival Control
Proprietary predictions, deltas, attention signals, evidence records and performance metrics.
●Licensed sources
Third-party aviation data is intended to be used under written rights appropriate to the production scope.
●No raw-data resale
Customers buy THOR outputs. THOR is not designed as a resale channel for supplier raw feeds.
●No provider proxy
Customer systems do not gain unrestricted access to supplier APIs through THOR.
●Customer truth first
Where available, client AODB/FIDS data provides schedule, identity and operational ground truth.
●Hosted in the EU
The THOR platform and its database run in Germany (Hetzner). Each customer only ever sees its own airports.
●Non-safety use
THOR Flights is for ground-resource planning and arrival readiness, not ATC, navigation or safety-of-flight decisions.
Start free. Pay when THOR has proven itself.
Indicative prices excluding VAT. The final price is set on the order form, according to station, volume and integrations.
- THOR on your own arrivals
- Up within 30 min of approval
- No commitment, no data to send
- Shadow mode against your baseline
- Weekly measures P2 → P3 → P4
- Deductible from the subscription
- One station, web space and API
- Attention queue and evidence
- Monthly proof report
- Several stations or a network
- Integrations (AODB/FIDS)
- Dedicated follow-up
Don’t believe the demo. Measure THOR on your traffic.
A Proof Sprint runs THOR in shadow mode against the customer’s own baseline and ground truth before any operational dependency is required.
Define truth
Selected site, flights, baseline, success criteria, data access and operational questions.
Run in parallel
THOR produces predictions and evidence while existing operations remain unchanged.
P2 → P3 → P4
Prediction performance, changed attention/actions, then measurable time and financial value.
Go or no-go
Convert only if the agreed evidence supports operational value and an annual deployment.
Frequently asked questions
What exactly does THOR predict?
The time each arriving flight will reach your airport, continuously over the last two hours before landing, together with the change against plan, its trend and an explicit attention state.
Where does the data come from?
From qualified commercial aviation data used under contract, and, when available, from your own operational data (AODB/FIDS). You receive THOR outputs, never the suppliers’ raw feeds.
How accurate is THOR?
On a full week of real traffic at Nice (636 flights, 10–17 October 2026): a mean error of 3.3 minutes 15 minutes before landing (98% of arrivals within ±10 minutes), 3.5 minutes at 30 minutes, 9.0 minutes one hour before. Two hours before, the mean error is 23 minutes: THOR is an arrival-readiness tool, not a timetable. Figures are measured against observed landing times, published per airport from 30 measured flights, and sealed every week. During your demo you also see, flight by flight, the time announced by the airline next to THOR’s and the actual landing.
What does the free demo include?
Seven days of THOR on the airport of your choice, in a private demo space, with your real arrivals. It starts within 30 minutes once your request is approved. No commitment and no data to send.
What is a Proof Sprint?
Eight to twelve weeks in shadow mode on your station, measured against your own baseline with criteria agreed in advance. From €10,000 excl. VAT, deductible from the subscription if you continue.
How do we use it day to day?
In a web space (screen, tablet, wall display), through an API, or by CSV export. Integration with your AODB or FIDS is optional.
Can THOR be used for safety decisions?
No. THOR is a ground-resource planning and arrival-readiness tool. It must never be used for air traffic control, navigation or safety-of-flight decisions.
Which airports can THOR follow?
Any airport of the reference table (about 29,000 airports, 4,500 with scheduled traffic in the search above). The showcase airports are followed live on this page.
Built by THOR 42 LAB.
THOR Flights is developed and operated by THOR 42 LAB, a French R&D and software company based in Paris.
Founded and governed by AVA Capital. AVA Capital reports 200+ founders supported, 100+ equity participations and €350M+ in financing rounds organised.
Request your free live demo — or a Proof Sprint.
Tell us which airport and which operation you want to test. Do not send sensitive operational files through this public form.