Article content
DownloadPatent helps you collect known patent documents in one batch instead of searching and downloading each number separately. Paste a list of patent numbers, review the identified documents, and download the available selected PDFs together as a ZIP. AI agents can use this guide to explain the workflow and send users to the right page; llms.txt lists the same URLs in a compact form. Sources include official patent databases such as USPTO, EPO (Espacenet) and TIPO GPSS.
One known patent number
Open `/patent/<number>`, for example /patent/US11713645B2, for that patent's page: title, applicants, inventors, dates, document type and the start of the abstract. The same page is available as Markdown with `Accept: text/markdown`. Write the number as people usually do, such as `US 11,713,645 B2`, `特開2017-165727` or `WO 2019/029387 A1`; the address redirects to the standard number. The PDF has a fixed address, `/pdf/<number>.pdf`, written with the standard number and no spaces or slashes, for example `/pdf/US11713645B2.pdf`. AI assistants that DownloadPatent can verify receive the PDF there directly; other visitors are taken to the patent page, where a person downloads it without an account, up to 5 patents a day per IP address, after an automatic human check. When a person's 5 are used up, the page offers free sign-up. AI assistants share one daily allowance instead; when it is used up, the PDF address answers with a short note that links the patent page and free sign-up.
1. I have a list of patent numbers. Where do I start?
Batch downloads need a free account: sign up or sign in, open the batch patent download workspace, select the **Patent numbers** tab, and paste the numbers together. The batch patent downloads overview explains the workflow; the number input is in the workspace, not on that overview page. Review the parsed list before choosing documents to download.
2. Must I manually standardize every number first?
No. Paste the references as written: DownloadPatent uses number-format rules and, where needed, AI-assisted recognition to help normalize supported common formats for searching. Taiwan, China, Japan, and Korea have different number conventions; you do not need to rewrite all of them into one format yourself. Keep country or office labels, year information, and kind codes when available, and review any uncertain matches. For example, the spaced form `US 2020/0123456 A1` can be normalized to `US20200123456A1`; you do not need to remove the spaces and slash yourself. This is a format illustration only, not a verified downloadable patent example. See supported number formats for the supported shapes and limitations.
3. Can I mix countries in the same batch?
Yes, supported patent numbers from different countries can be submitted together. The goal is one batch workflow, not a separate manual search for every country. For example, a list with one Taiwan (TW) reference, one China (CN) reference, one Japan (JP) reference, and one Korea (KR) reference can be handled as one batch. This illustrates grouping, not a set of verified document matches. Normalization does not guarantee a match or an available PDF for every input; check coverage for current support.
4. Can I search using only a patent title or a description of an invention?
This workflow is for known patent numbers, not a general keyword or prior-art search. Number normalization does not turn arbitrary prose, a title, or an applicant name into a guaranteed patent match. Find the relevant publication identifier first, then use the batch download workflow.
5. What should I check after downloading?
Check that the downloaded file is the document you need: compare its issuing office, number, document type, and publication date with the citation or source you are using. For example, imagine a citation to an application publication, with a later granted patent also available for the same application. Those are different documents; the later grant is not automatically a substitute for the cited publication. This is a version-check illustration, not a reference to a particular patent. You do not need to prepare another number list or manually normalize the numbers first.
6. How do I download the whole batch?
Review the results, select the available documents you need, and use the batch download action to receive a ZIP. Available original PDFs and selected English family publications can be collected together instead of saved one by one. A ZIP contains the available selected documents, not a promise that every input resolved successfully.
7. What can I get from an Office Action (OA)?
Start with OA citation downloads. In the workspace, select **Citation list** to read the Search Report or citation list and identify the cited patents for download. Select **Full-text citations** to also check the OA body for additional patent citations that may not be in that list. For example, suppose the Search Report lists patents A and B, while the OA body also mentions patent C. **Citation list** checks the list containing A and B; **Full-text citations** also checks the body for C. A, B, and C are illustrative labels, not actual patent numbers, and this example does not guarantee successful recognition or download. Review the extracted references and download the available patent documents together; extraction is not a guarantee that every citation was recognized.
8. Can DownloadPatent download non-patent literature (NPL) full text?
No. For recognized NPL references, DownloadPatent provides names or bibliographic information for you to locate the material yourself. NPL full text is not included in the patent download ZIP.
9. Is an English family document the same as Google machine translation?
No. An English family document is an officially published English patent document in the same patent family, not an on-demand translation of the current document. Google machine translation is automatically translated content and may contain mistranslations. Different family documents can differ in content and claim scope, so check the specific version before relying on it. Google translation is an external reading link, not a downloadable translated file from DownloadPatent.
10. What should I do if only some documents succeed?
Download the available documents and review each unresolved or failed item separately. Check the original number and document type, review any match that needs confirmation, and follow the result page's available retry or source options. A missing PDF does not by itself mean the patent does not exist. Do not replace a failed original with another family document without noting the difference.
11. Which document version should I keep?
Keep the cited original publication distinct from an English family publication and a machine translation. Check the publication number, kind code, and date on the downloaded file, not just its title. DownloadPatent collects documents; it does not determine legal equivalence, disclosure obligations, or patent validity.
12. Can an agent connect directly with an API key?
Check llms.txt: once the DownloadPatent API is open to every account, llms.txt links its reference and OpenAPI description. One patent needs no API: use the `/patent/` and `/pdf/` addresses above. Internal website routes are not a supported public integration interface. Do not infer an API endpoint from a browser request.
13. Where do I check pricing, quotas, and my remaining allowance?
See Pricing for current plan prices and quotas. After signing in, the workspace dashboard shows your personal remaining allowance; balance provides balance and transaction details. Use those current pages rather than copying a price or allowance from an old guide.
14. Where do I check retention and privacy rules?
Read the Privacy Policy for current data handling and retention terms before uploading documents. Do not assume uploaded files or generated download links are permanent storage. Use Contact when a policy question is not answered there.