Octa Models creates realistic environments of real places: the foundational ground on which any operation on Earth is planned, built and run.
Modelled, not drawn. Every operation answers to the forces of its place: the rock and soil beneath it, the water moving through it, the air and climate around it, the sun above it. These ecological forces weigh more on a decision than any plan. We model them storey by storey on one octagon of land, from the authorities' own data, to the half metre where the country has mapped it by LiDAR. Every value names its source. Where none exists, you see a dash, not a guess.
Breda–Montseny, Catalonia · 29 July, 18:55 · the Ground model at the golden hour, the sun setting behind the Montseny
Fifty-seven seconds inside Octa Models: Barcelona’s ground from the authorities’ own LiDAR, every building one click away, the Besòs flowing to the sea, the air over the city, every equation shown — and every star above the land.
LiDAR, aerial photo, geological map: ICGC (CC BY 4.0) · buildings: D. G. del Catastro · air: ERA5 (Copernicus) · rivers: OpenStreetMap, HydroSHEDS · music: original · voice: the founder · more in the Newsroom
Octa Models builds a living model of one octagon of ground, from named sources, and opens it to you in 2D and 3D, one plain question per tab.
A map shows where. It rarely shows why. What is known about a place sits with dozens of agencies, in different formats and scales. Octa Models puts their answers on the same ground.
A fruit of Octa Intelligence: it grows from its modelling engine and applies the eight aspects to one exact place.
Each model is a storey of the same ground, built on the one beneath it: water lies on the rock, air moves over the water, and the sky turns above all of them. Every tab opens with the question it answers.
Four steps, the same for every place. You pay for the modelling once, then a monthly subscription for the model and the data on it.
Click the map to set its centre and choose its size. You see its area and price at once.
Ground, Water, Air and Sky. Tick a whole model or only the items you need.
We gather the data for your exact octagon from named sources, build the model and check it by hand.
Explore it in 2D and 3D, tab by tab. Every value shows its source and quality.
We’d rather show a dash than guess. A place without data shows a dash and a grey surface, never another place’s numbers.
Read from a named official source.
From a named non-official source, labelled as such.
Checked by people, at the desk or on site — quoted case by case.
We measured the demo at Breda–Montseny against independent references — national survey data and NASA’s own ephemeris — not against itself. These are the results.
a pixel at the closest view, from Catalonia’s own LiDAR survey (ICGC, flown 2021–2023, at least 8 points a square metre).
In everyday termsAbout ten height readings along one parked car — enough to see a terrace wall, a ditch or a stream bank.
how far the free world elevation model (Copernicus, 90 m) sits from Breda’s true ground, typically — more than 10 m at nearly a quarter of the octagon, up to 40 m. The demo reads the national survey instead.
In everyday termsLike measuring a field and getting it a two-storey house too high — in places a twelve-storey block too high. On a flat valley floor, that is enough to make water seem to run uphill.
from NASA JPL’s ephemeris on average, over every daylight hour of 2024; never more than 0.22°.
In everyday termsA 10 m tree’s shadow falls within 15 cm of where NASA puts it, and a sundial would read within a minute of it.
every tab of the four Ecological models opened in 3D, in 2D and as a table before launch — with no error.
In everyday terms87 screens, each opened and checked, the way a pilot runs the checklist before take-off.
how far Catalonia’s 25 cm aerial photo (ICGC, 2025) sits from the national LiDAR, measured on four squares 2 km across: 0.5 m in the valley, up to 1.6 m on the wooded slopes, where tree crowns lean in a photo taken from the air.
In everyday termsOne stride. Each pixel is the size of a dinner plate, and a tree in the photo stands within a step of where the laser found it.
of Breda’s 15,998 building parts stand at the height the laser measured (ICGC LiDAR, surface minus bare ground); the rest at their cadastral floors × 3 m, and each card says which.
In everyday termsA one-storey house stands about 5 m to its ridge, a two-storey 7 m, a three-storey 10 m — as measured, not as guessed.
Where two sources disagree, each does the job it does best. At Breda the 25 km air record gives about a third more rain than the 4 km climate record, which is corrected by the weather stations — so the model takes how much it rains from the climate record, and when it rains, hour by hour, from the air record. Every tab names the source it read.
Measured on the demo, 3–4 October 2026 · Breda–Montseny, an octagon 11.1 km across · sources: ICGC Territorial LiDAR · Copernicus DEM GLO-90 · NASA JPL Horizons · ERA5 · TerraClimate
Place the octagon, set its size, tick the models. The area is what you pay by. Above 300 km across, we’ll quote you.
Every tab of every model opens in three forms at once — a searchable table, a 2D map and a 3D model — over the same values, the same moment and the same sources. Move between them in a fraction of a second: nothing reloads, nothing is lost, and the item you picked stays picked.
Opens in 0.08 sEvery number behind the picture, as rows you can read, compare and take away.
Ready in 0.4 sThe ground laid flat on a real map, every feature drawn true to where it is.
60 frames a secondYour ground as a block you can turn in your hands, lit by the real sun of the hour you choose.
No settings pages, no waiting: every choice redraws the model at once, in all three forms.
Two buttons on every tab open the engine itself: what it computed, and what it cost to compute.
Behind every tab runs Octa Intelligence: one engine that takes a centre and a size anywhere on Earth and builds the same storeys on that ground — the rock, the water, the air and the sky — from named sources, by published equations, in your browser. It reads every dataset from its own store, works out millions of values in moments, and keeps the table, the map and the block in step with every click.
It is fast because it is built to be: the heavy work runs beside the page, the ground sharpens only where you look, and what it has read once, it keeps. It is trustworthy because it is built to be: every value names its source, every equation its reference — and where there is no source, it shows a gap instead of a guess.
Measured at the show place, Breda–Montseny, on a laptop (Apple M1 Max), 3 October 2026. The equations, datasets and files are counted from the engine at its last build.
From the first request to the open model: the invoice, the payment, the build and the check, each step shown as it happens — and the model one click away when it opens.
Whatever you run — a solar plant, a wind farm, a battery park, a data centre — its meters can send their readings to its model with one plain web request. They show within a minute on the real ground, under the real sun: on each part’s card, in each part’s colour and on the plant’s own dashboard. Move the clock and they replay.
Made in your account in a click. It can only send readings in: it never reads Octa’s data or changes your account.
Name each one — an inverter, a turbine, a charger, a soil probe — with its place on the ground and what it measures. Tie it to a part of your layout and its readings show on that part.
Every minute, or in batches from the monitoring system you already run: plain HTTPS and JSON. Sending the same reading twice never makes a double.
Each part lights in its live colour, its card shows its values and a chart of the day, and the plant’s dashboard adds them up: power, energy, alarms, earnings.
All your projects
Solar plant
Its dashboard108 tables in 18 rows facing south at 35°, 10.4 m apart so no row shades the next in winter, laid inside one parcel of the land registry, 5 m within its boundary, each row as long as the parcel allows; 6 inverters and a 2 MWh battery.
Every table, every inverter and combiner box, the sunlight meter in the panels’ plane, the battery, the transformer and the grid meter.
148 meters · 1,187 values, every minute
Wind farm
Its dashboard3 × 4.2 MW, 136 m rotors on 112 m towers, their substation and a met mast.
Each turbine’s power, wind, rotor speed and yaw; the met mast; the substation’s transformer, switchgear and grid point.
9 meters · 109 values, every minute
Battery park
Its dashboard8 battery containers, 4 inverter skids and a 50 kV transformer on a gravel yard, the solar plant behind.
Each container’s charge and cell temperatures, the inverters, the transformer, the grid meter, the fire-water tank.
16 meters · 124 values, every minute
Green hydrogen plant
Its dashboardTwo 1.25 MW modules, compressors behind fire walls, 600 kg stored at 500 bar, a refuelling station.
The electrolysers, the water treatment, the compressors, the storage, the trailer bays, the gas detectors and the station.
15 meters · 111 values, every minute
Data centre
Its dashboardTwo data halls on two power feeds, UPS rooms, coolers and four hydrogen fuel cells for back-up.
The halls’ load and temperature, the UPS, the coolers, the fuel cells, the batteries, the network and the grid.
19 meters · 122 values, every minute
Water supply
Its dashboardFour wells in the valley’s gravel, a treatment plant, two tanks on the heights and a small hydro to run the pumps.
The wells’ levels and flow, the treatment’s quality, the pumps, the tanks’ levels and the hydro’s power.
15 meters · 133 values, every minute
EV charging hub
Its dashboard4 × 150 kW and 6 × 50 kW beside the main road, a 117 kWp solar canopy and a 200 kWh buffer battery.
Each charger’s power and session, the canopy, the buffer battery, the hub’s transformer and meter.
14 meters · 117 values, every minute
Smart farm
Its dashboardEight fields of vegetables, barley, maize and wheat, each with its soil probe; a well, a tank and a weather station.
Soil moisture in every field, rain and weather, the well and the tank, the water each field took, the panels over field 3.
13 meters · 128 values, every minute
POST /api/v1/places/{place}/readings
Authorization: Bearer okm_live_…
{"readings": [{"source": "inv-01",
"t": "2026-10-04T12:00:00Z",
"v": {"ac_kw": 310.5}}]}
The eight are Octa’s own examples at Breda–Montseny, each placed by the site finder (how the ground is chosen): they are not real plants, and their readings are simulated — sent through the same API yours would use. Pictures: the 3D model on Catalonia’s 25 cm aerial photo (ICGC 2025, CC BY 4.0), 15 September at 13:40, true height.
The model is not only read — it is used. See the ground photographed, bring your own operation onto it, and ask Claude to work the model for you in plain words.
Two views beside Relief, Grey, Wire and Satellite, under every model.
The national mapping agency's own aerial photograph, laid on your ground and kept in your model.
Google's photographed 3D — every building, tree and street as photographed — cut to your octagon.
Where Google has no 3D city, the authorities' own data stand the buildings up.
The laser survey, the infrared aerial photo and the land-cover map are grids of millions of cells. Octa's tools read them and draw every tree patch, hedge, meadow, field and building as a shape with its area, perimeter and height — clickable, in a table, ready to export.
Breda–Montseny, 4 October 2026: ICGC LiDAR 2021–2023 and infrared aerial photo 2025, the ICGC land-cover map 2024 and the national cadastre; patches under 50 m² (trees, shrubs) or 200 m² (meadows, crops) folded into their surroundings.
What you send stays yours, drawn in its own violet beside our data — never mixed with it.
Upload a site plan or a survey and it appears on the model, in 2D, 3D and the table.
Inverters, sunlight meters, weather masts, soil probes, gauges: send their readings and they show live on the model.
Claude inside your modelPress Ask beside the model. Claude reads what is on your screen, turns the model — the tab, 2D or 3D, the clock — reads the values at any spot, your layers and your readings, takes you there, and answers with the figures and where they come from.
At 4 pm on 21 December, which of my panel rows are in shade?Claude moves the clock to that hour, reads your panel table with the sun and the shade on each row, and flies you to the rows in shade.
How deep is the water table under my site in August, compared with March?It opens Groundwater, sets the two months on the clock, reads the depth at your site in each, and gives you both with the difference.
Which meadows larger than 2 hectares lie within 500 m of the river?It reads the meadows' table and the rivers, lists the patches with their areas, and takes you to the first one.
Did inverter 3 make less power yesterday than the sunlight allowed?It reads your live readings for that day beside the model's own sun for each hour, and points to the hours that fell short.
Where are the tallest trees near the north fence?It switches the canopy on, reads the tree patches and their heights along your fence, and frames them in 3D.
Present the survey I just uploaded.Suggest with Claude proposes a name, the colours and the columns worth showing. Nothing changes until you accept it.
Your question and what is on your screen go to Anthropic (Claude) to be answered; Octa's code and equations never do. Each answer is billed by what it uses and shows its price. See the prices →
Tell us the place and what you need to know about it. We answer with the models that fit, and a price.
Octa Intelligence develops contextual and spatial intelligence through data infrastructure, modelling, mapping, analysis and simulation. Octa Models brings it to one exact place.