Google Patents Scraper: Search, Claims & Citations

Search Google Patents across the USPTO, EPO, WIPO, CNIPA, JPO, KIPO and 100+ other offices, or look up patents by number, and get clean JSON: title, abstract, assignee, inventors, priority / filing / publication / grant dates, anticipated expiration, legal status, CPC classifications, claims, cited patents, cited-by, family members, legal events, images and the PDF link. No API key, no Google account. Proxy (datacenter, with automatic residential fallback) is included in the price.

Two prices, and you only pay for what you ask for:

• Search records (patent-found, $1.50 / 1,000): the result list, fast. Number, title, snippet, assignee, inventors, all dates, family country status and the PDF link (granted patents).
• Full records (patent-detailed, $3.70 / 1,000): everything on the patent page. Used for patent-number lookups and for search results when Full details is on.

What you can do with it

• Prior-art search: run keyword searches with the same syntax as patents.google.com, then pull claims and citations for the hits.
• Competitor patent portfolios: every patent of an assignee (Waymo, Apple Inc), optionally narrowed by keyword, CPC class, country or date.
• IP landscaping and technology trends: filter by CPC code (G06N3/08, H01M10/0562) and priority date, sort by newest.
• Patent monitoring: schedule a daily or weekly run sorted by newest to catch new filings in your field.
• Citation analysis: forward citations (cited-by) and backward citations (prior art), with examiner-cited flags.
• Citation networks and families: start from patent numbers and get the full record of every patent they cite, every patent citing them (up to 3 hops) and every family member filed at other offices.
• Expiry and legal-status checks: look up a list of patent numbers and get status, expiration date and legal events.

Input

Field	What it does
searchQueries	Keyword searches (solid state battery, "neural network" AND pruning) or pasted patents.google.com search URLs.
patentIds	Patent numbers or URLs for full records: US9876543B2, US-10000000-B2, EP3000000B1, WO2017120318A1.
assignees, inventors, cpc	Filters for every search. Each also works alone (an assignee's whole portfolio).
countries	Patent offices: US, EP, WO, CN, JP, KR, ...
dateFrom, dateTo, dateType	Date range on priority, filing or publication date.
status, patentType, languages	Granted vs applications, utility vs design, original language.
sortBy	relevance, newest, oldest.
includeDetails	Turn search hits into full records (charged as patent-detailed).
followCitations, citationDepth	For patentIds: also return full records of the patents they cite (backward), that cite them (forward) or both, 1-3 hops. Each says relation, relatedTo and networkDepth.
includeFamily	For patentIds: also return full records of every family member.
includeClaims, includeDescription	Claim text (on by default) and the full specification text (off by default; large).
detailLanguage	en (English or Google's translation) or the original language, e.g. de, ja.
maxItems	Hard cap on results for the whole run (and on cost). Default 50.

Search for new US grants on solid-state batteries:

{ "searchQueries": ["solid state battery"], "countries": ["US"], "status": "GRANT",
  "dateFrom": "2023-01-01", "sortBy": "newest", "maxItems": 200 }

A company's lidar portfolio with full details:

{ "searchQueries": ["lidar"], "assignees": ["Waymo"], "includeDetails": true, "maxItems": 100 }

Look up specific patents:

{ "patentIds": ["US9876543B2", "EP3000000B1", "https://patents.google.com/patent/CN110573898B/en"] }

Citation network around one patent (everything it cites and everything citing it, two hops, at most 200 records):

{ "patentIds": ["US9876543B2"], "followCitations": "both", "citationDepth": 2, "maxItems": 200 }

Output

A search record:

{
  "publicationNumber": "US20260133317A1",
  "title": "Distance-velocity disambiguation in hybrid light detection and ranging devices",
  "snippet": "FMCW lidars utilize high precision technology and are complex and expensive devices. A less expensive option can be to use a hybrid lidar device ...",
  "assignee": "Waymo Llc",
  "inventors": ["Bryce Remesch"],
  "priorityDate": "2021-08-04", "filingDate": "2025-12-22", "publicationDate": "2026-05-14", "grantDate": null,
  "countryCode": "US", "language": "en",
  "familyCountryStatus": {"US": "ACTIVE"},
  "pdfUrl": null,
  "url": "https://patents.google.com/patent/US20260133317A1/en",
  "searchRank": 2, "searchQuery": "lidar", "searchTotalResults": 8250, "detailed": false
}

A full record adds (shortened):

{
  "publicationNumber": "US9876543B2",
  "title": "Systems and methods for channel state information exchange",
  "abstract": "Systems and methods are disclosed herein to provide improved channel estimation ...",
  "assignee": "Keysight Technologies Singapore Sales Pte Ltd",
  "currentAssignees": ["Keysight Technologies Singapore Sales Pte Ltd"],
  "originalAssignees": ["Keysight Technologies Singapore Holdings Pte Ltd"],
  "inventors": ["Thomas Alexander", "Lester Noel Stott"],
  "priorityDate": "2016-01-05", "filingDate": "2016-01-05", "publicationDate": "2018-01-23",
  "grantDate": "2018-01-23", "expirationDate": "2036-01-05", "legalStatus": "Active",
  "kindCode": "B2", "applicationNumber": "US14/988,707", "familyId": "59235955",
  "cpcCodes": ["H04B7/0417", "H04B7/0617", "H04B7/0626", "H04L1/1664", "H04L1/1671"],
  "cpcClassifications": [{"code": "H04B7/0417", "description": "Feedback systems", "first": true, "inventive": false}],
  "claimsCount": 20, "independentClaimsCount": 16,
  "firstClaim": "1. A system for transferring a channel estimate for wireless digital communications, comprising: ...",
  "claims": [{"number": 1, "text": "1. A system for ...", "dependent": false}],
  "citationsCount": 6,
  "citations": [{"publicationNumber": "US20050259686A1", "priorityDate": "2004-05-07", "publicationDate": "2005-11-24",
                 "assignee": "Michael Lewis", "title": "Initiation of communication", "examinerCited": true}],
  "citedByCount": 4, "citedBy": [{"publicationNumber": "US11108442B1", "assignee": "Qualcomm Incorporated", "...": "..."}],
  "familyCitedByCount": 14,
  "nonPatentCitations": ["Paul et al., \"Wireless LAN Comes of Age ...\" (2008)."],
  "alsoPublishedAs": [{"publicationNumber": "WO2017120318A1", "publicationDate": "2017-07-13"}],
  "similarDocuments": ["US20170195016A1", "CA2856049C"],
  "events": [{"date": "2016-01-05", "title": "Application filed by ...", "type": "filed", "critical": true}],
  "legalEvents": [{"date": "2016-02-25", "code": "AS", "title": "Assignment", "attributes": {"Owner name": "IXIA, CALIFORNIA"}}],
  "imageUrls": ["https://patentimages.storage.googleapis.com/.../US09876543-20180123-D00000.png"],
  "pdfUrl": "https://patentimages.storage.googleapis.com/ad/75/30/1bcbbfcb1519af/US9876543.pdf",
  "externalLinks": {"espacenet": "https://worldwide.espacenet.com/...", "globalDossier": "https://globaldossier.uspto.gov/..."},
  "language": "en", "translatedFrom": null,
  "url": "https://patents.google.com/patent/US9876543B2/en",
  "detailed": true
}

Machine-translated pages (CN, JP, KR, DE...) return the English text only; translatedFrom names the original language. A run summary (per-search counts, Google's total match count, failed inputs) is saved as SUMMARY in the key-value store.

Pricing

Pay per result, no start fee beyond Apify's own:

• $1.50 per 1,000 search records (patent-found)
• $3.70 per 1,000 full records (patent-detailed)

A patent is charged once, at one of the two prices, never both. Filters and maxItems apply before you are charged. Cheaper on Apify's higher plans. Proxy cost is included.

Limits and behaviour

• Google Patents serves at most 1,000 results per search. For more, split by date range, country or CPC class.
• Requests are sequential and paced (~1.5 s apart) so Google does not throttle the run: about 40 full records or a few thousand search records per minute. If Google rate-limits even after switching proxy IPs, the run stops and keeps what it already has.
• A run that finds nothing fails with a clear message instead of returning an empty dataset. Unknown or mistyped patent numbers are listed in SUMMARY.failed and the rest of the run continues.
• Data is what Google Patents shows; legal status, expiry and current assignee are Google's estimates, not legal conclusions.