French DVF API comparison 2026: 10 ways to access French property prices

There is no single best DVF API — the right choice depends on what you cross-reference with the prices. For a prototype or a one-off need in a single commune, Etalab's DVF API is free and sufficient. To explore French open data from an assistant, data.gouv.fr's official MCP server is free and needs no key. To combine transactions with energy ratings and building data, or to plug an AI agent into specialised tools, Normi is the only one of the ten solutions compared here that brings DVF, ADEME energy diagnostics and the CSTB building database behind a single API. For multi-country coverage, PriceHubble serves several European markets whereas Normi covers France only.

DVF (Demandes de Valeurs Foncières) is the French tax administration's public record of every property sale since 2014, published by the DGFiP under an open licence. It is the closest French equivalent to the UK Land Registry's Price Paid Data or US county deed records: actual transacted prices recorded before a notary, not asking prices from listings. DVF goes further than Price Paid Data in one respect — it includes the floor area, the number of rooms and the cadastral parcel reference, which makes a price per square metre computable per transaction. It is more limited in another: Alsace-Moselle uses a separate land-register system, so coverage there is partial.

Yes, several. Etalab's DVF API (api.data.gouv.fr/dvf) is free and needs no key. The CEREMA's DVF+ files are freely downloadable. data.gouv.fr's official MCP server is also free and unauthenticated. These public sources genuinely cover prototyping and one-off analysis, and there is no reason to pay while they suffice. What they do not provide is geocoding, a normalised price per square metre, or cross-referencing with other datasets — that preparation work is what commercial solutions charge for.

Not quite, for a specific technical reason. The official data.gouv.fr MCP server exposes nine read-only tools that search datasets, read their metadata and query resource contents through the Tabular API. That Tabular API caps resources at 100 MB per CSV file, while the complete annual DVF files exceed that limit. The server is therefore excellent for discovering and exploring the French open-data catalogue, DVF included, but it is not designed to serve analytical queries across the full transaction record. It is free, official and keyless: it is the right place to start.

Models range from free to enterprise contracts. The public sources (Etalab, CEREMA DVF+, data.gouv.fr MCP) are free. Commercial APIs with public price grids sit, as of 9 September 2026, between roughly thirty and a hundred euros per month: Immo API lists €29–49/month, ImmoData plans start around €49/month. Enterprise vendors such as PriceHubble and Yanport quote on request. Normi runs on usage-based credits with 500 free credits each month and no card required, which suits intermittent usage better than a flat subscription. All of these prices change without notice.

Of the ten solutions compared here, Normi is the only one exposing all three behind a single API: DVF transactions, ADEME energy performance diagnostics matched by cadastral reference, and the CSTB's BDNB building database. ImmoData advertises DPE endpoints on its developer portal. The others are limited to transactions, or list the ADEME and BDNB datasets in a catalogue without providing the match to the sold property — and that match is precisely the expensive part.

Yes, through MCP (Model Context Protocol), which lets an assistant call tools with no integration code. Three of the solutions compared here expose an MCP server: data.gouv.fr's (official, free, nine read-only catalogue tools, Streamable HTTP transport only), Immo API's connector, and Normi's (over thirty specialised tools across DVF, DPE, BDNB and SIRENE, over both Streamable HTTP and STDIO). For the other solutions, no public MCP server was identified as of 9 September 2026.

They all draw on the same DGFiP publication, so maximum freshness is identical everywhere: nobody can hold data more recent than the official release. Since 2026 the DGFiP publishes annually (previously twice a year), and current data covers the full 2025 calendar year. There is therefore a lag of several months to a year between an actual sale and its appearance in DVF. For market-trend analysis that lag is negligible; for real time you need listings data, which DVF does not contain at all.

Automated valuation needs geolocated comparables, which means geocoding and a normalised price per square metre — something the public sources do not provide as-is. Among the solutions compared, PriceHubble and Yanport offer an AVM as their core product, enterprise-oriented and quoted on request, with PriceHubble covering several European countries. Normi exposes geolocated comparables and a low / mid / high valuation range through its REST API and MCP tools, on usage-based credits, for France only.

Yes. DVF is published under the Etalab Open Licence 2.0, which permits reuse including for commercial purposes, provided the source is attributed. This is why several commercial APIs can legitimately build products on the same underlying public dataset — what differs between them is not the raw data but the preparation, the enrichment and the access model. Buyer and seller identities are anonymised in DVF, so it carries no personal data of that kind.