French Sold House Prices (DVF): Sale Price, Surface & Price per m²

Every real French property sale, straight from the official land registry. Give it a commune name, postal code or INSEE code, and you get back structured transaction records: sale price, address, property type, surface, rooms, land, and price per square meter (€/m²), pulled from the government's Demandes de Valeurs Foncières (DVF) open data.

These are sold prices, not asking prices — the actual figure written into the deed. Built for property valuation and comparables (comps), real-estate market analysis, proptech, investment sourcing, and AI agents (works via API and MCP).

Coverage — checked 2026-08-09

Years available	2021, 2022, 2023, 2024, 2025 (verified live against files.data.gouv.fr/geo-dvf; 2026 is not published yet)
Geography	Metropolitan France + most DOM, every commune
Paris / Lyon / Marseille	Auto-expanded to arrondissements — Paris or 75011 both work
Lookup keys	Commune name, 5-digit postal code, INSEE code — mixed freely in one run
Run reliability	29/29 successful runs in the last 30 days, 0 failed, 0 timed out

Why this actor

• ✅ Official data. This runs on the DGFiP/Etalab geo-DVF dataset, so every row is a real recorded sale, not an estimate or a listing.
• ✅ Query how you think. Search by commune name (Bordeaux), postal code (75011) or INSEE code (33063), and Paris, Lyon and Marseille get auto-expanded to their arrondissements.
• ✅ Price per square meter computed for you, along with a clean address, geocoordinates, surface, rooms and land area.
• ✅ Real filters: property type, price range, minimum surface, standard sales only, and year (2021 to 2025).
• ✅ Deduplicated. One sale can span several lots and parcels; rows are deduplicated by mutation, parcel, type and surface so your comparables are not double-counted.
• ✅ Fair pricing. You pay per transaction returned, $1.00 per 1,000, platform usage included. A search that matches nothing costs you the start fee only.

Use cases

• Estimate a property's value from comparable nearby sales — build a comps set filtered by property type, surface and area, and read the price per square meter.
• Market analysis that tracks French house prices, price trends and sale volumes by commune, arrondissement or department.
• Investment sourcing across multiple areas at once, filtered by budget and surface.
• Enrichment, attaching real transaction history to an address or parcel.
• AI agents can query it over MCP, asking Claude or Cursor things like "average €/m² for 2-room flats sold in Bordeaux in 2024?"

Input

{
    "locations": ["Bordeaux", "75011", "33063"],
    "years": ["2023", "2024"],
    "propertyTypes": ["Appartement", "Maison"],
    "minPrice": 100000,
    "minSurface": 30,
    "onlySales": true,
    "maxResults": 1000
}

locations accepts commune names, postal codes and INSEE codes, mixed freely.

Every input

Input	Type	Default	What it does
locations	array	(required)	Commune names, postal codes or INSEE codes, mixed freely. Paris, Lyon and Marseille expand to their arrondissements.
years	array	most recent year	Which DVF years to read (2021 to 2025). Leave it empty and you get the latest year only.
propertyTypes	array	["Maison", "Appartement"]	Local type as DVF names it: Maison, Appartement, Dépendance, Local industriel. commercial ou assimilé.
minPrice	integer	none	Drop sales below this price (€).
maxPrice	integer	none	Drop sales above this price (€), which is how you keep an outlier château out of a flat's comps.
minSurface	integer	none	Drop sales under this built surface (m²).
onlySales	boolean	true	Keep only standard sales ("Vente"), dropping VEFA, exchanges and expropriations.
maxResults	integer	1000	Hard ceiling on transactions returned, and therefore on what a run can cost you.

Output (one record per transaction line)

{
    "source": "DVF",
    "date": "2024-01-04",
    "year": 2024,
    "nature": "Vente",
    "price": 557500,
    "currency": "EUR",
    "pricePerM2": 10519,
    "propertyType": "Appartement",
    "surfaceBuilt": 53,
    "rooms": 3,
    "landSurface": null,
    "address": "12 RUE DE LA ROQUETTE",
    "postalCode": "75011",
    "commune": "Paris 11e Arrondissement",
    "inseeCode": "75111",
    "department": "75",
    "parcelId": "75111000AB0123",
    "latitude": 48.856,
    "longitude": 2.376,
    "sourceRef": "files.data.gouv.fr/geo-dvf (Etalab / DGFiP)"
}

Companion actors in the French real-estate suite

DVF gives you the sold prices (what buyers actually paid). Pair it with live asking prices from the listing scrapers to see the gap:

• SeLoger Scraper, for asking prices from France's #1 portal, with seller phone plus particulier/agence flag.
• Logic-Immo Scraper pulls Logic-Immo asking prices, with the advertiser's phone number and agency/private flag.
• PAP Real Estate Listings covers owner-direct (particulier) asking prices from PAP.
• Real Estate Deal Score combines PAP asking prices with this DVF data to rank listings over or under market.

Pricing

Pay per event, no subscription:

Event	Price	When
Transaction found	$0.001 ($1.00 / 1,000)	Per transaction returned
Actor start	$0.001	Once per run

$1.00 per 1,000 transactions, all in

Platform usage (compute, proxies) is included in the price — nothing is billed on top. A 1,000-row comps pull costs $1.00, or $0.70 on the Gold plan (Bronze, Silver and Gold discounts apply to every event). Set a per-run charge cap if you want a hard ceiling, and note that a run returning nothing costs you the $0.001 start and no more.

FAQ

Where does the data come from? The official geo-DVF dataset published by Etalab from DGFiP records (files.data.gouv.fr/geo-dvf). It covers metropolitan France and most of the DOM; Alsace-Moselle notary data is not in DVF.

How recent is it? DVF is published twice a year; this actor serves years 2021 through 2025 (checked 2026-08-09 — 2026 is not published yet). Sales can take a few months to appear after signing.

Sold prices or asking prices? Sold. Every figure is the price recorded at the deed, which is why DVF is what valuers and lenders use. For asking prices, pair it with the listing scrapers below.

Why several rows for one sale? A single mutation can span multiple lots/parcels, so each line is one property component. Records are deduplicated by mutation, parcel, type and surface.

Do Paris/Lyon/Marseille work? Yes, a name or postal code for these cities gets automatically mapped to the correct arrondissement(s).

Does it work with AI agents? Yes, it's exposed via REST API and MCP.

---

Something missing?

If you need an extra field, another source, or a different output, open an issue on this Actor and describe it. I read every request and small additions usually ship within days. More French real estate Actors (pige, DVF, deal score, DPE leads) are on my profile.

Keywords: DVF, valeurs foncières, French sold house prices, price per square meter France, French property prices, property transactions France, real estate comparables France, prix immobilier, prix au mètre carré, €/m², estimation immobilière, transactions immobilières, French land registry, cadastre, proptech, DGFiP, Etalab.