EPO Espacenet & Open Patent Services
Espacenet is the European Patent Office's public patent search service, offering free access to over 140M patent documents from more than 100 countries. Every record carries bibliographic data (titles, applicants, inventors, classification codes, and citations), and for a large share of them the full text and drawings as well. Alongside the familiar web interface, Open Patent Services (OPS) exposes the same data through a programmatic API, and the EPO's text-analysis tooling has made getting at patent full texts far less painful than it once was.
For anything past casual browsing you will want OPS. You register for a free account, receive API credentials, and query the service, which answers in `XML` with some JSON. The free tier carries weekly quotas on how much you can pull, so plan any bulk ingestion deliberately: decide the corpus you actually need, fetch it, cache it, and stay inside the allowance rather than hammering the endpoint and getting throttled.
As a RAG source it is excellent for reading, searching, and analysing patents: prior-art searches, mapping a technology landscape, or grounding a technical assistant in genuine invention disclosures. Patents are dense, highly structured, and heavily cross-referenced, which suits passage retrieval well, and the international spread means you are not boxed into a single jurisdiction.
The access model is the thing to get right, and it is why we rate this as limited. The EPO's terms let you use the data freely for most purposes but stop short of granting a blanket right to reuse it inside a commercial product: think of it as "open access, not open licence." You can read the documents at no cost, yet building a product on top of them is restricted, so if you intend to commercialise anything derived from Espacenet, check the EPO's conditions or arrange a specific licence first. Coverage varies too, since full text is not available for every country or every era, and the documents span many languages.
WIPO PATENTSCOPE gives similar global reach through a different interface, the USPTO publishes open bulk US patent data, and Google Patents Public Datasets on BigQuery is worth a look when you need large-scale, query-friendly access.
Related sources
BIGPATENT
A corpus of 1.3 million US utility patents filed between 1971 and 2018, each paired with its human-written abstract as a gold-standard summary and organised by Cooperative Patent Classification code. A large, clean patent text corpus built for abstractive summarisation and other patent NLP work.
Google Patents Public Datasets
A query-based collection of over 120M patent documents from more than 100 patent offices worldwide, including applications, pre-grant publications, and granted patents. Accessible through Google BigQuery, with hundreds of millions of USPTO events also queryable.
Harvard USPTO Patent Dataset (HUPD)
A large-scale, structured corpus of US patent applications built specifically for machine learning and natural language processing research. It fills the gap left by mainstream patent search tools, which are not designed with the ML and NLP community in mind.
PatentsView
A platform from the US Patent and Trademark Office (USPTO) for exploring US patent data, with visualisation and analysis tools. It connects patents, inventors, organisations, and locations in a regularly updated database, and offers both an API and complete bulk data tables.