Infrared Core

The climate simulation enginePhysics simulation and AI surrogate models, fused on a scalable backend. Wind, felt heat, sun and daylight for any site, from one room to a whole city, in seconds.

Buildings, trees and terrain © Lands Department, HKSAR.

37 s100 km² of Hamburg

12 modelsWind, heat, sun, energy

0.1 to 1 mGrid, rooms to cities

Agent-readyHeadless, for AI agents

The same simulations everywhere

One engine behind every tool, so every result is comparable.

Infrared Core. One engine for felt heat, wind, sun, solar, daylight and energy, in our cloud or on your servers.

On demand. API and SDK, Grasshopper, QGIS, ArcGIS, the browser platform or your own AI agent.

Live layer. Your city or twin prepared once, about 500x faster, as map tiles and a queryable dataset.

Enterprise readyHosted in the EUYour data stays yoursOn-prem possibleSLA available

A spin-off of the Austrian Institute of Technology, built with our technology partners

Scalable infrastructure

A whole city runs at once

From one room to a whole city: Infrared Core splits any area into tiles, runs them in parallel in our cloud or on your servers, and merges one seamless result.

Any size. A block, a district or a whole city, with no fixed limit.
In parallel. 100 km² takes 37 s, not 100 times as long as 1 km².
Reliable. A failed tile is sent again. No holes, no double bill.
Headless. Ready for AI agents: they call the same API and read the results.

REST API
Python and TypeScript SDK
AI agent skill
3D Tiles and GeoTIFF

From a room to the whole city

We run it in the cloud, at every scale. The same engine gives daylight in a room, sun on each facade, heat in a street and a map of a whole territory.

Daylight on each floor of one tower in Wan Chai — RoomInterior daylightDaylight factor, floor by floor.

Solar radiation on every facade of a dense block in Wan Chai — FacadeSun on every facadeRadiation on walls and roofs.

Perceived heat across a district of Hong Kong Island — DistrictHeat across a districtStreet by street, at 1 m.

Perceived heat over 65 by 49 km of Hong Kong, computed in tiles and merged — TerritoryA whole territory65 × 49 km, in tiles, merged.

Hong Kong. Buildings, trees and terrain © Lands Department, HKSAR.

Twelve models, one engine

The same buildings, trees and weather go in. Choose the question: wind, comfort, heat and cold, sun, daylight or energy.

Wind speed: an example result — **Wind speed**m/s at pedestrian level

Pedestrian wind comfort: an example result — **Pedestrian wind comfort**Lawson, Davenport, NEN 8100

Thermal comfort: an example result — **Thermal comfort**Time in UTCI comfort

Heat stress: an example result — **Heat stress**Hours above the UTCI threshold

Cold stress: an example result — **Cold stress**Hours below the UTCI threshold

Microclimate (UTCI): an example result — **Microclimate (UTCI)**Perceived heat (UTCI) in °C

Daylight availability: an example result — **Daylight availability**Direct-sun score

Sky view factors: an example result — **Sky view factors**Visible sky in %

Direct sun hours: an example result — **Direct sun hours**Hours of sun

Solar radiation: an example result — **Solar radiation**kWh/m² on roofs and facades

Interior daylight factor: an example result — **Interior daylight factor**Daylight in a room

Energy balance: an example result — **Energy balance**Heating and cooling need

The SDK guide counts ten analyses, not twelve: thermal comfort, heat stress and cold stress are three models here and one SDK analysis, thermal comfort statistics.

All 12 models in the simulation library

100 times the area in 37 seconds

Large areas are split into tiles. The tiles run in parallel in our cloud and come back as one grid. So 100 km² takes 37 s, not 100 times as long as 1 km².

When a result takes seconds, you can ask "what if we move the tower six metres?" in the meeting and see the answer.

1 km²Vienna · 6,904 buildings · geometry already uploaded 5 s

9 km²Vienna · 28,714 buildings 24 s

100 km²Hamburg · 59,883 buildings 37 s

400 km²Berlin · 318,856 buildings 44 s

Perceived heat (UTCI) for one summer day, 08 to 18 h, on a 1 m grid, on our production cloud. 1 km²: geometry already uploaded, measured 24 Oct 2026 (first run with upload about 17 s). 9 to 400 km²: end to end, building data and upload included, measured 23 Oct 2026. Berlin ran as four 100 km² runs at the same time.

Tested against Radiance, EnergyPlus, field instruments at four sites, and CFD. See the validation

Looking at ENVI-met, SOLWEIG or PALM-4U? See how infrared.city compares with ENVI-met, SOLWEIG, PALM-4U and Orbital Stack