Google Lens API

Google Lens reverse image search, products and OCR, as JSON

endpoints
4
credits per call
0–2

Google Lens reads the text in a picture and translates it.

/google-lens/v1

  • POST/google-lens/v1/ocr2 credits
  • POST/google-lens/v1/translate2 credits
  • POST/google-lens/v1/objects2 credits
  • POST/google-lens/v1/languages0 credits
4 active endpoints, on 0 and 2 credit tiers.Every parameter ↓OpenAPI 3.0 spec ↗

2 credits per call≈ $2/1,000 calls1,000 free credits, no card

Get a free API key
Try it live

Run the call yourself, on the real endpoint

Change a parameter and press Try it. The response on the right is what your code will receive.

Google Lens API

4 of 4 endpoints, ready to run

View docs ↗

Every paragraph, line and word in a picture, with boxes and language

2 credits0 required · 1 optional
POST/google-lens/v1/ocr
idle
// Press "Try it" and this pane shows exactly what the
// live site returned this second — including an empty
// result, if that is the truth. No key, no account.

Try every endpoint in the playground

Endpoints

What Google Lens endpoints does ReefAPI ship?

4 live read endpoints. Read-only data API: no writes, no account actions, no dashboard access on the target site.

  • ocr

    2 cr

    Read all text in a picture with Google Lens.

    required
    —
    optional
    image_url, image_base64, region
  • translate

    2 cr

    Translate the text in a picture with Google Lens into one of 249 languages, paragraph by para…

    required
    target_language
    optional
    image_url, image_base64, source_language, region, include_background_images
  • objects

    2 cr

    The regions Google Lens detects in a picture.

    required
    —
    optional
    image_url, image_base64
  • languages

    0 cr

    The languages Google Lens translates from and into (codes + names), as Lens itself lists them.

    required
    —
    optional
    display_language

Every parameter, every allowed value →

Missing a Google Lens endpoint, or need a source we don't have yet?Contact us real people · same-day reply.
How it works

How the Google Lens API works

Google Lens is a normal ReefAPI surface — the same four rules that hold for every other engine on the key.

Authenticate01
x-api-key header

No OAuth app, no request signing, no per-site account. One key covers all 746 engines.

Call02
POST /google-lens/v1/…

Every route is a POST with a JSON body. Parameters are validated against the published schema before anything is charged.

Pay03
0 or 2 credits per call

Credits, not seats. Failed and blocked calls are never charged, and nothing is cached: every call reads the source live.

Read04
{ ok, data, meta, error }

One envelope everywhere. meta carries latency_ms, record_count and the endpoint that answered.

Recipe

From a photo to a translated, positioned text layer

Read a picture, translate it, and place each translated line where the original was.

019780140328721-L.jpg
GET/google-lens/v1/ocr?image_url=https://covers.openlibrary.org/b/isbn/9780140328721-L.jpg

The Fantastic Mr Fox cover: 4 paragraphs, 5 lines, 12 words, language en, with a box for each and 4 detected regions.

02languages
GET/google-lens/v1/languages

The 249 target languages with their names.

039780140328721-L.jpg&target_language=de
GET/google-lens/v1/translate?image_url=https://covers.openlibrary.org/b/isbn/9780140328721-L.jpg&target_language=de

"Der fantastische Mr. Fox" split into the cover's two lines with the box of each original line and the colours to draw them in; "ROALD DAHL" kept as a name.

A translated text layer you can draw over the original picture, line by line.

request
curl -X POST https://api.reefapi.com/google-lens/v1/ocr \
  -H "x-api-key: $REEF_KEY" \
  -H "content-type: application/json" \
  -d '{"image_url":"https://covers.openlibrary.org/b/isbn/9780140328721-L.jpg"}'
response envelope
{
  "ok": true,
  "data": { … },
  "meta": {
    "api": "google-lens",
    "endpoint": "ocr",
    "mode": "live",
    "latency_ms": …,
    "record_count": …
  },
  "error": null
}
Fields

What a Google Lens field actually contains

Fields that read differently from how they look. Every row was measured against live Google Lens on 2026-11-19.

FieldWhat it holds
boxcenter_x, center_y, width, height as fractions 0-1 of the picture, x / y = the top-left corner, rotation_rad = clockwise rotation in radians. `pixels` gives the same box in pixels when the picture's size can be read from the file (JPEG, PNG, WebP, GIF, AVIF, BMP).
separatorWhat follows a word on its line: a space, or nothing ("MR" + "." in "MR. FOX"). Joining text + separator rebuilds the line exactly.
languageThe language Lens detects: one for the whole picture and one per paragraph (ja for the Kabukichō gate sign, ru for "Привет мир", el for "Καλημέρα κόσμε").
status (translate)Per paragraph: success; no_op_translation when Lens keeps the text as it is (names such as "ROALD DAHL"); same_language when the paragraph is already in the target language; unknown_source_language; empty_translation. translated_text of the whole picture keeps the original text of paragraphs Lens did not translate.
start / end (translate lines and words)Character positions in the translated paragraph: the Japanese line "ファンタスティック・" is 0-10 of "ファンタスティック・ミスター・フォックス".
text_color / background_colorThe colours Lens uses to redraw a translated line on the picture, as #RRGGBBAA.
background_imageFormat, size and paddings of the patch Lens paints behind a translated line; with include_background_images=true also the patch itself (WebP, base64).
objects[].kindsalient_region = the main subject, detected_object = a separate object, whole_image = the full frame. Lens gives no name or label for a region.
region (parameter)Keeps the words whose centre lies inside x,y,width,height. Lens still reads the whole picture; it is a filter on the result, not a crop.
matches[].price (visual_matches, products)The price Google prints on the match card, as printed ("$5*", "3 €*") plus amount and currency. The asterisk is Google's own mark. Checked against each card's own label on 2026-11-19: every parsed price equal.
matches[].rating / review_countThe star rating and the count in brackets that Google prints on the card (AbeBooks "4.1 (153,627)"). Google does not say on the card whether they rate the product or the store. Empty on most non-shop cards.
country (results actions)The market Google answers for: asked for de, a US English book cover came back with medimops.de offers in EUR. It changes stores, prices and currency, not the picture's reading.
exact_matches[].google_linkGoogle's own redirect to the page. It opens the page from about 3 minutes after the call; before that Google shows a "Verifying your request" wait page (measured 0 of 12 links at 0-120 s, 12 of 12 at 180 s).
image_width / image_height (exact_matches)The size of the copy of the picture on that page, as Google prints it (678x1,000).

Lens's language and country settings did not change the text read from a picture (en-US, ja-JP, tr-TR and de-DE gave identical results), so ocr and translate have no locale parameter. The results-page actions do take country and language: they change the matches and prices.

Coverage

Google Lens coverage, measured

Every number below was read from live Google Lens on 2026-11-19: 82 calls over the 4 reading actions, 82 answered, 13 checks that wrong inputs get the right error, and a SOAKHOURS-hour soak of SOAKCALLS results-page calls.

Text accuracy

6 of 6 pictures whose text we wrote came back exact (PNG, JPEG, WebP, GIF, AVIF); Cyrillic and Greek 2 of 2; every title word found on 8 of 8 book covers from 8 subjects; the Kabukichō gate photo read "歌舞伎町一番街" (ja).

Translation

249 target languages. 29 of 29 translate calls answered, e.g. "Donde esta la biblioteca" → "Where is the library", "Bonjour tout le monde" → "Herkese merhaba", "Привет мир" → "Hello World". Status per paragraph; names are kept as written.

Regions

2-4 regions per photo (main subject, separate objects, whole frame) with boxes; no labels.

Speed

Median 1.5-2.9 s per call for ocr, translate and objects, 0.4 s for languages; slowest call 5.4 s; base64 pictures 0.9-1.3 s. One picture per call.

Reverse image search

SOAKLINE

Limits

Pictures up to 20 MB. Results-page actions read JPEG, PNG and WebP (AVIF and GIF uploads got no matches, so they are refused). Exact-match page links open from about 3 minutes after the call. A question about the picture needs a Google sign-in and is not offered. Lens's language and country settings do not change the text read; `region` filters the OCR result, it does not crop.

Use cases

What people build with Google Lens

The jobs this data is most often used for.

01

Read text from photos, screenshots, signs, menus and book covers.

02

Translate a menu, sign or label photo into the reader's language, line by line.

03

Redraw a picture with translated text using Lens's own line boxes and colours.

04

Find where the text and the main subject sit in an image before cropping or tagging it.

4

endpoints

0/2

credits per call

Pricing

What Google Lens data costs

The cheapest call here is 0 credits, so $15/mo (Pro) buys 10,000 of them — $1.50 per 1,000 credits. Credits roll over and never expire, and failed or blocked calls are not charged.

$0.67–$1.50 / 1,000 credits
Full pricing →
  • 1,000 free credits on signup, no card
  • One key, all 746 APIs, one credit pool
  • Failed and blocked calls are never charged
  • Credits roll over and never expire
no credit card1,000 free credits · instant API key · pay by card or crypto
Quickstart

Call it in two lines

Sign up, get 1,000 credits and one key that works on every engine. Then this is the whole protocol.

curl
curl -X POST https://api.reefapi.com/google-lens/v1/ocr \
  -H "x-api-key: $REEF_KEY" \
  -H "content-type: application/json" \
  -d '{"image_url":"https://covers.openlibrary.org/b/isbn/9780140328721-L.jpg"}'
python
import os
import requests

REEF_KEY = os.environ["REEF_KEY"]

r = requests.post(
    "https://api.reefapi.com/google-lens/v1/ocr",
    headers={"x-api-key": REEF_KEY},
    json={
        "image_url": "https://covers.openlibrary.org/b/isbn/9780140328721-L.jpg"
    },
    timeout=90,
)
print(r.json()["data"])
FAQ

Have a question? We got answers.

The questions people actually ask before wiring up Google Lens.

Which picture formats work?▾

JPEG, PNG, WebP, GIF and AVIF were read live with the same text from the same picture. Give a public URL or the file as base64 (a data: URI also works), up to 20 MB.

Can I give a web page instead of a picture?▾

Yes. When the address is a page, the picture the page declares as its preview (og:image) is read: a GitHub repository page gave its 1200x600 preview card. A page that declares no picture returns INVALID_PARAM.

How accurate is the text?▾

On 6 pictures whose text we wrote ourselves, Lens returned the exact text 6 times, including PNG, JPEG, WebP, GIF and AVIF files; Cyrillic and Greek text 2 of 2. On 8 book covers from 8 subjects every word of the book's title was found.

Which languages can it translate into?▾

249, the list Lens itself publishes; the `languages` action returns it with names in the language you choose. Use a code (de, ja, zh-CN) or the English name (German).

Why is part of the picture not translated?▾

Lens leaves names and text already in the target language as they are; each paragraph carries a status (no_op_translation, same_language) that says so. "ROALD DAHL" stays "ROALD DAHL" in German.

Does it return visual matches and similar products?▾

Yes. `visual_matches` returns the pages Google Lens shows for a picture (up to 60, in Google's order) with page link, title, site and the matched image; `products` keeps those with a printed price, in the currency of the country you ask for. A book cover gave 60 matches, 18-20 of them priced.

Can it find where a picture is used on the web?▾

Yes, `exact_matches` lists the pages that carry the exact picture: 191 to 399 per picture in our tests, with page title, site, date, the size of the copy and Google's link to the page. Google's link opens the page from about 3 minutes after the call.

Can I ask a question about the picture?▾

Not yet: Google answers a picture-plus-question search only to signed-in users, and this API does not sign in. `ai_overview` returns Google's own AI description of the picture with its sources.

Does it name the objects it finds?▾

No. `objects` returns where the regions are (the main subject, separate objects and the whole frame), not what they are; Lens does not send a label for them.

How fast is it?▾

ocr, translate and objects: median 1.5-2.9 seconds in 82 live calls (languages 0.4 s), slowest 5.4 s. Results-page actions in the SOAKCALLS-call soak: median SOAKP50 seconds, 95 % within SOAKP95 seconds. When many calls arrive at once the results actions can answer 429 "warming up — retry shortly" within 5 seconds; retry after 5 seconds.

What is the Google Lens API?▾

Google Lens API is a ReefAPI endpoint group for google lens as json: reverse image search (the pages and products that show a picture, with prices), the pages that carry the exact picture, google's ai answer about it, and the text in the picture read and translated into 249 languages. It returns live JSON through POST requests under /google-lens/v1.

Is the Google Lens API free to try?▾

Yes. ReefAPI starts with 1,000 free credits, no card required. Google Lens calls use the same shared credit balance as every other ReefAPI engine.

Do I need a Google Lens login or account?▾

No login to Google Lens is needed for the API response. You call ReefAPI with your x-api-key header, and the playground can run live examples before you create a production key.

How fresh is the Google Lens data?▾

The page example is captured from a live ocr call, and production requests fetch live data through ReefAPI rather than a static sample.

Same key

19 Utilities & AI APIs on the same key

One key, one credit pool, one response envelope. If you are pulling Google Lens, you are one call away from the rest of the category — no second contract, no second integration.

Contact

Need something this API does not do?

Name the endpoint, the field, or a source we do not carry yet. We ship new APIs every week and you would be first to get the key. Real people read every message and reply the same day.

0/4000

No account needed · we reply from [email protected]

Try it on your own data before you pay anything

The call above is the real endpoint, not a recording. A free key gives you 1,000 credits, the other 745 APIs, and the same envelope everywhere.

Endpoints, parameters and credit costs on this page are read from the live catalog and cannot drift from what the API accepts. Field notes were captured on 2026-11-19.