Google Cloud Vision API and Document AI
Google's cloud OCR: Vision for text in images, Document AI for structured extraction.
Opens the vendor's own site.
Jobs covered
Open a task to see exactly what the vendor documentation supports.
OCR
Browse all ocr toolsdetecting and extracting text with Google Cloud OCR, where the documentation separates Cloud Vision text detection on images and on PDF and TIFF files, including online synchronous small batch annotation of PDF, TIFF and GIF files and the TEXT_DETECTION and DOCUMENT_TEXT_DETECTION features, from Document AI Enterprise Document OCR, which detects and extracts text and layout information from PDFs, scanned document images and Microsoft DocX files with configurable image quality score, language hints and rotation correction, and optional add-ons for math formulas, font style information and checkbox extraction that require Enterprise Document OCR version 2.0 or later, with Cloud Vision text detection described as a globally available REST API with a default quota of 1,800 requests per minute
Where your files are processed
Reported to upload your files
The vendor states that files are uploaded to their servers. We have not independently tested this.
Documents are uploaded to Google Cloud. Online requests are processed in memory rather than persisted, while batch requests store the file briefly and delete it after processing.
There is no self-hosted deployment: the only on-premises-style option is documented as deprecated.
What it does
Two Google-hosted, consumption-billed services at different levels, which is why they share one entry. Cloud Vision is an annotation API whose OCR modes return text and layout - for sparse text in an image, or structured pages for dense documents.
Document AI is a processor platform: enterprise OCR returns text and layout with deskew and rotation correction, while separate processors add form parsing, invoices, expenses, bank statements, and custom classifiers and splitters that return key-value pairs and entities. Neither returns a redacted document.
Best for
Developers who need high-volume text or layout extraction from images and PDFs, or teams that need typed form fields, business-document entities and document classification.
Worth knowing
Two products sit behind this listing and their limits are not the same. Cloud Vision takes images up to 20 MB and PDFs up to 1 GB; Document AI is 40 MB online and 1 GB in batch, with page limits that depend on the processor.
An eligible imageless online request can span 30 contiguous pages from page one instead of the 15 usually quoted. The linked data-usage notice is Vision's and does not establish Document AI retention or processing behaviour, and Document AI values do not apply to Vision.
Neither is a redaction tool.
Details, and where each comes from
Vendor statements, our own reading of their pages, and the things nobody has established yet, each answered separately. Where we have not confirmed something it says so rather than guessing, and a row nobody has established is greyed out.
- Versions covered
- Web · API
- Website
- cloud.google.com
- Requires an account
- Yes
- Free tier
- Yes
- Watermark on the free tier
- No
- Maximum file size
- 40 MB
- Batch processing
- Yes
- Data residency
- Multiple regions
- Open source
- No
- Self-hostable
- No
- Licence
- Not verified
- Redaction
- No redaction feature
- Programmatic access
- HTTP API
- Last availability check
- Responded in 159 ms 21 October 2026
Checked automatically by our own runner, not reported by the vendor. 100% of our last 30 checks got a response.
What the redaction verdict rests on
From the vendor's own documentation, read 22 September 2026. Not tested by us.
What that does not establish
The vendor’s own site was read on 22 September 2026 and contains no mention of redaction, sanitisation, blackout or whiteout. Absence from a vendor’s own material is not proof that a feature does not exist - it may be undocumented, or on a page the reading did not reach - but for a NEGATIVE claim it is the strongest evidence available, and it is weaker than a positive quote.
Recorded as such rather than presented as certainty.
What the vendor says
Claims are labelled rather than repeated as fact. Where we have checked something ourselves, it says so and shows when. Checked means we read the vendor's own page and it says what we recorded - not that we tested the behaviour, unless the note says we did.
-
“Pricing is tiered - the first 1000 units used each month are free, units 1001 to 5,000,000 are priced as marked, etc.”
Checked by us - the source says this Read 22 September 2026
Confirmed verbatim: the sentence introduces the pricing table exactly as recorded. The table behind it gives the rest of the structure and is worth carrying, because "etc." in the claim is doing real work: the tiers are the first 1,000 units a month free, units 1,001 to 5,000,000, and units 5,000,001 and above, each with its own price per 1,000 units.
The page also notes that the final 1,000-unit block is prorated, so a month that ends mid-block is not billed as a whole one. Recorded as the vendor's published pricing; no API call was made and none was billed.
Standing: Re-read and matched byte for byte
Check it yourself: the source, the SHA-256, and how we checked
- From
- the text a browser renders
- SHA-256
- 4d201a7f9fcc4a80a087ea29113b801dea85b73a0cfee9c0f5687696ec25df12
How we checked
Re-verified verbatim on 2026-09-22: https://cloud.google.com/vision/pricing re-rendered in headless Chrome with scripts/render-page.mjs and the claim's wording found in the rendered text. Originally verified on 2026-09-21. sourceHash is the SHA-256 of the re-rendered text, held as rv-cloud-google-com-vision-pricing-688562fd.txt.
-
“Google does not use the content you send to train and improve our Cloud Vision features such as its machine perception model.”
Checked by us - the source says this Read 22 September 2026
Confirmed verbatim: the sentence appears unaltered on the data-usage page, which is the right page for it - a vendor's training-data commitment belongs in documentation rather than in marketing, and this is documentation. The claim is specific in a way that matters: it says Google does not use the content "to train and improve our Cloud Vision features such as its machine perception model", which scopes the commitment to that product's models rather than to Google generally.
A reader should not read it as a general statement about their data. Recorded as the vendor's documentation; no API call was made.
Standing: Re-read and matched byte for byte
Check it yourself: the source, the SHA-256, and how we checked
- From
- the text a browser renders
- SHA-256
- 1dfeb57189fd13e35fe2cbf3f1781841658c20427d239cb16dc95b933f36954e
How we checked
Re-verified verbatim on 2026-09-22: https://cloud.google.com/vision/docs/data-usage re-rendered in headless Chrome with scripts/render-page.mjs and the claim's wording found in the rendered text. Originally verified on 2026-09-21. sourceHash is the SHA-256 of the re-rendered text, held as rv-cloud-google-com-vision-docs-data-usage-5aa43ced.txt.
Each claim is quoted from the vendor's own pages. This section describes what we have and have not checked - it is not a judgement about anyone.
Sponsorship is distribution, not an endorsement or a promise of search rankings, AI citations, traffic, or sales. This listing is a free entry, compiled from published sources.
Visit Google Cloud Vision API and Document AI Report a problem