Get the weekly radar
One short email with the freshest products, every week.
Floor Plan AI turns one written sentence into a 2D layout or a downloadable 3D model in seconds.
Grounded in available product and source data
No CAD software or draftsperson is usually needed just to test five different room layouts before calling an architect — that's specifically what Floor Plan AI, a text-to-floor-plan and 3D model generator, skips, turning one written sentence into a layout instead.
A description like bedroom count, rough size, and a must-have feature is enough to run either of two modes: 2D generates a full top-down layout with walls, doors, and labelled rooms in 20-60 seconds, while 3D Visualization, still in beta, produces a downloadable GLB model in 2-3 minutes that opens directly in Blender, Unity, Unreal, three.js, or Apple Vision Pro. Seven drawing styles, from technical blueprint to watercolor to photorealistic, cover different end uses without switching tools.
A layout doesn't have to start from a blank prompt either — up to 16 reference photos, whether sketches, existing plans, or interior shots, can be uploaded so the AI restyles a space while keeping its underlying structure intact, useful for iterating on something already partly decided rather than generating from scratch each time.
The site names three specific audiences its output is built for: homeowners comparing renovation layouts before bringing one to a professional, real estate teams generating listing-ready plans and interactive 3D walkthroughs without a draftsperson, and 3D designers importing a generated GLB straight into an existing rendering or game pipeline, with commercial use included on paid tiers. Google sign-in is required to generate, and the first 1K-resolution plan is free before any payment decision.
No — a plain-English description of the space (or the template tab's clickable room counters) is the input; no CAD vocabulary or manual drafting is required to get a first layout.
Yes, through 3D Visualization specifically, which is marked beta — it produces a downloadable GLB model that opens in Blender, Unity, Unreal, three.js, or Apple Vision Pro, separate from the faster 2D top-down mode.
Yes — image-to-image references let up to 16 photos (including existing sketches or plans) be uploaded, and the AI restyles the layout while keeping the underlying structure a user already has in mind.
No — the site states it does not train models on user designs and does not share user inputs with third parties.
Yes — Google sign-in is required, but the first 1K-resolution plan is free, letting a layout be generated before any decision about upgrading has to be made.
