WIPO PATENTSCOPE
PATENTSCOPE is the free patent search service run by the World Intellectual Property Organization (WIPO), the United Nations agency for intellectual property. Its defining strength is international reach: one interface searches applications filed under the Patent Cooperation Treaty, the route inventors use to seek protection in many countries at once, alongside national and regional collections from more than 100 patent offices. A typical record carries the full text, claims, abstract, applicant and inventor details, classification codes, and often the drawings.
Two features make it stand out for anyone working through patents at scale. Cross-lingual search lets a query in one language retrieve documents published in others, which matters because patents appear in dozens of languages. The built-in machine translation, tuned specifically on patent wording, then lets you read a Japanese or Chinese filing in English well enough to judge whether it is relevant. You search and read through the website; there is no wholesale bulk-download product, so it is best understood as a discovery and reading layer rather than a corpus you ingest in one pass.
For RAG, PATENTSCOPE suits prior-art searches, freedom-to-operate research, and technical question answering where you need authoritative patent text and the broadest possible cross-jurisdiction coverage in a single place. When your pipeline needs machine-readable bulk data to index directly, pair it with the download-friendly collections: PatentsView and Google Patents Public Data offer US and global full text you can query or export in volume, with PATENTSCOPE providing the international breadth they may miss.
The caveat to weigh is licensing. Access costs nothing, but reuse is governed by WIPO's own terms rather than a clear open licence, so what you may lawfully do with the retrieved text beyond searching and reading is genuinely unclear. Confirm WIPO's conditions before you store, redistribute, or build a commercial product on the content, and do not assume free access implies free reuse.
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.
EPO Espacenet & Open Patent Services
Free access to over 140M patent documents from the European Patent Office (EPO). Its Open Patent Services text analysis tools have made patent full texts much easier to reach, though the terms are open access rather than an open reuse licence.
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.