Google Patents Scraper - Claims, Citations & Legal Status

Patent details from Google Patents by publication number: title, abstract, assignees, inventors, priority, filing, publication and grant dates, legal status, CPC codes, claims, citations, family ID and PDF link. One flat JSON record per patent.

What it does

• Takes a list of patent publication numbers (any patent office covered by Google Patents: US, EP, WO, CN, JP, DE, KR, GB, FR...) and returns one record per patent.
• Reads only the public patent page https://patents.google.com/patent/<number>/<language>, one request per patent.
• Accepts numbers with or without kind code (US7650331B1, US7650331), with spaces or punctuation (US 7,650,331 B1, US2004/0267807 A1), or full Google Patents URLs.
• Returns the claim count and the first claim for every patent; the full claims, the full description and the list of cited patent numbers are optional.
• English text by default (Google machine translation for non-English patents), or the original language, or another language supported by Google Patents.
• Duplicate inputs are fetched once. Numbers that do not exist on Google Patents (HTTP 404) are reported in the run status and not charged.

Typical uses: enriching a patent portfolio or a list of numbers exported from another tool, prior-art and citation analysis from known patents, monitoring legal status and assignees.

Input

Parameter	Type	Required	Default	Description
patentNumbers	array of strings	Yes	—	Publication numbers or patents.google.com/patent/... URLs
includeClaims	boolean	No	false	Add the text of every claim (claims)
includeDescription	boolean	No	false	Add the full description (description), often several hundred KB
includeCitations	boolean	No	true	Add the publication numbers cited by the patent (citations)
language	string	No	en	en, original, de, fr, es, it, pt, nl, ja, zh, ko, ru

Example:

{
    "patentNumbers": ["US7650331B1", "EP1234567B1", "WO2020123456A1"],
    "includeClaims": false,
    "includeCitations": true,
    "language": "en"
}

Output

One dataset item per patent (real output, abstract and first claim shortened):

{
    "publicationNumber": "US7650331B1",
    "countryCode": "US",
    "kindCode": "B1",
    "title": "System and method for efficient large-scale data processing",
    "abstract": "A large-scale data processing system and method includes one or more application-independent map modules ...",
    "assignees": ["Google LLC"],
    "currentAssignees": ["Google LLC"],
    "inventors": ["Jeffrey Dean", "Sanjay Ghemawat"],
    "applicationNumber": "US10/871,244",
    "priorityDate": "2004-06-18",
    "filingDate": "2004-06-18",
    "publicationDate": "2010-01-19",
    "grantDate": "2010-01-19",
    "status": "Active",
    "expirationDate": "2026-12-23",
    "cpcCodes": ["G06F9/5066", "G06F9/544"],
    "claimsCount": 16,
    "firstClaim": "1. A system for large-scale processing of data, comprising: ...",
    "claims": null,
    "description": null,
    "citations": ["US3905023A", "US6192359B1", "US6182061B1"],
    "citationsCount": 18,
    "citedByCount": 323,
    "familyCitationsCount": 0,
    "familyCitedByCount": 0,
    "nonPatentCitationsCount": 6,
    "familyId": "41509957",
    "pdfUrl": "https://patentimages.storage.googleapis.com/7c/97/eb/f084e6ce6281d8/US7650331.pdf",
    "inputNumber": "US7650331B1",
    "language": "en",
    "url": "https://patents.google.com/patent/US7650331B1/en",
    "scrapedAt": "2026-11-02T10:00:00+00:00"
}

Field notes:

• grantDate is the grant date of the same application (also filled for an A1 publication whose application was later granted); null if not granted.
• status and expirationDate are Google's estimate, shown on the page with Google's own disclaimer that they are not a legal conclusion.
• citationsCount / citedByCount count patent-to-patent citations of this publication. Some offices (often EP) only have family-level citations: see familyCitationsCount and familyCitedByCount.
• A field is null (or an empty list) when Google Patents does not show it for that document. On a 50-patent sample across 21 offices, title, dates, application number and family ID were always present; the abstract was missing on 14 of 50 (mostly older or JP/GB/Nordic documents), the claims on 8 of 50 and the legal status on 13 of 50.

Pricing

Pay-per-event, no start fee: $1.40 per 1,000 patents (event PATENT) on the Apify Free plan, lower on paid plans. You only pay for results pushed to the dataset.

Apify plan	Price per 1,000 patents
Free	$1.40
Starter	$1.20
Scale	$1.00
Business	$0.84

Set a maximum cost per run in the run options: the Actor stops cleanly when it is reached.

Limitations

• No search. The robots.txt of patents.google.com is Disallow: /* followed by Allow: /patent/ (checked on 2026-11-02). Only patent pages are allowed to crawlers, so this Actor does not query the Google Patents search (/?q=, /xhr/query) and never sends a request to it. Input is a list of patent numbers you already have.
• One request per patent: a patent's citations are returned as numbers; to get their details, run the Actor again with those numbers.
• Text in English is Google's machine translation for non-English documents.
• Data is what Google Patents displays; it is not an official register. Legal status, assignees and dates may differ from the patent office's register.
• Inventor names are part of the public bibliographic record and are returned as published.

Related Actors