Changelog - cloro
Changelog
Changelog
Stay up to date with the latest cloro API product updates, new features, enhancements, bug fixes, and announcements about our monitoring endpoints.
Subscribe to the changelog’s RSS feed at
https://cloro.dev/docs/changelog/rss.xml
in any feed reader, or
add it to Slack
to get live updates in a channel.
Navigate to changelog: 19th November 2026
Google Search: hotel cards inside text ads are now in
ads
Some text ads show a carousel of hotel cards, often 20 per ad. The
Google Search endpoint
did not return these cards. Each card is now an item in
ads[]
:
{
"position"
:
1
,
"blockPosition"
:
"bottom"
,
"type"
:
"SHOPPING_CARD"
,
"category"
:
"Sponsored"
,
"title"
:
"DoubleTree by Hilton London ExCeL"
,
"url"
:
"https://www.google.com/aclk?sa=L&ai=..."
,
"page"
:
1
,
"price"
: {
"value"
:
84
,
"currency"
:
"£"
,
"raw"
:
"£84"
},
"store"
:
"Hilton"
,
"rating"
:
4.3
,
"reviews"
:
"3.3k"
}
blockPosition
is the block of the text ad, and
store
is its advertiser. See
Sponsored ads
.
Navigate to changelog: 19th November 2026
ChatGPT: what each inline product card is
ChatGPT shows brands, companies, software, places and ideas in the same cards as shop products. Each product in
inlineProducts
now has
kind
and
entityCategory
.
kind
is
product
for a shop product,
entity
for a named entity such as a brand, and
row
for any other card.
entityCategory
is the category that ChatGPT gives to the entity in the card title, such as
brand
or
tv_show
, or
null
. To keep only shop products, filter on
kind: "product"
.
price
now holds only a price. Before, a card for a business could have its rating or address in
price
. That line now stays in
subtitle
or
description
.
Navigate to changelog: 19th November 2026
ChatGPT: call-to-action links on product cards
A gpt-6 product card can show a call-to-action line, such as “Acheter chez TennisPro”. Each product in
inlineProducts
now has
ctaLinks
, with the label and URL of each merchant link on that line. Use it to tell these links apart from citation chips (
citationPillIds
) and product sheet offers (
offers
). A product with no call-to-action line has
ctaLinks: []
.
url
does not change.
Navigate to changelog: 17th November 2026
Google Search: expand the AI Overview in the HTML page
You can now expand the AI Overview in the
include.html
page. Send
include.aioverview.expand: true
and the stored page shows the full AI Overview, as if
Show more
were clicked. This works on desktop only: on phones, Show more opens AI Mode.
result.aioverview
does not change.
Navigate to changelog: 16th November 2026
Google Search: resolving redirect links is free for paying customers
POST /v1/monitor/google/goto
now costs 0 credits if you have an active subscription or bought credits. Before, it cost 1 credit per resolved link.
When Google changes how it serves its links, responses can have an empty
link
for some hours. You can now backfill those links at no cost: send each
redirectLink
to the endpoint.
Navigate to changelog: 16th November 2026
Google Search: 17 more SERP features
The
Google Search endpoint
now returns 17 more blocks that Google shows on the results page. A key is omitted when its block is not on the page. No flag, no extra cost.
| Key | Block |
|---|---|
| sportsResults | Standings, games, squads and stat leaders |
| relatedWebsites | Carousels of site cards (“Related websites”, “Find results on”) |
| inlineImages | The Images block |
| videos | The Videos carousel, with key moments |
| shortVideos | The Short videos carousel |
| weather | Current conditions, hourly and daily forecast |
| latestPosts | An entity’s own social posts |
| financeQuote | Market summary of a stock, fund, index, future or crypto currency |
| events | Event cards and ticket offers |
| discussionsAndForums | Forum threads, with their answers |
| thingsToKnow | The “Things to know” facets |
| transitResults | Train and bus routes, departures and fares |
| flights | The flights blocks, with prices, stops and booking sites |
| hotelResults | The hotel pack and hotel carousels, with prices and booking sites |
| stores | The merchants of the Stores block |
| comparisonShoppingSites | In the EEA, the products of each comparison shopping site |
| topSights | The top sights of a city or place |
Organic results also have a
readMore
link when Google ends the snippet with “Read more”.
Navigate to changelog: 12th November 2026
Perplexity is discontinued
Perplexity blocks access for logged-out sessions. Each workaround we found stopped working within days, so we discontinued the endpoint.
POST /v1/monitor/perplexity
and async tasks with
taskType: "PERPLEXITY"
return
400
(
MODEL_UNAVAILABLE
) and cost no credits. See
unsupported providers
.
Google Search: responses about 2x faster
The
Google Search endpoint
now returns a 1-page response in about 1.7 seconds at the median, against about 3.7 seconds before 7 November. These times are for requests without
include.html
and
include.aioverview
.
Navigate to changelog: 12th November 2026
Google AI Overview: product cards with merchant offers
The
Google Search endpoint
now returns the product cards in an AI Overview answer as
result.aioverview.inlineProducts
, with title, price, store and rating. A card that links to a Google product viewer has a
productResultToken
: send it to
POST /v1/monitor/google/product
to get the merchant offers. See
AI Overview inline products
.
Google AI Mode
inline products
now also have
rating
.
Navigate to changelog: 9th November 2026
Google: EU places unit in
localResults
In the EU, Google often replaces the local pack with a list of suppliers (Google, TheFork, UberEats, …). The
Google Search endpoint
now returns the places of every supplier in
result.localResults
, each with a new
supplier
field. Places also gain
ratingScale
(5 or 10) where Google shows one. No flag, no extra cost.
Navigate to changelog: 7th November 2026
New endpoint: Bing AI Mode
POST /v1/monitor/bing/aimode
returns the answer text, sources and citation pills from Bing AI Mode. Set
include.markdown
or
include.html
to also get the answer in Markdown or a URL to the rendered page.
Price : 4 credits, the same as Google AI Mode. A state adds +2 credits, and sync requests add the +2 sync surcharge .
Countries : all except CN , KR and RU , where Bing AI Mode does not answer. GET /v1/countries?model=bing_aimode returns the list.
Prompt length : 3,500 characters after URL encoding, so fewer for non-Latin scripts.
Async : use task type BING_AIMODE with POST /v1/async/task or POST /v1/async/task/batch .
Copilot is discontinued
Copilot blocks access for logged-out sessions, so we discontinued the
/v1/monitor/copilot
endpoint. Earlier Copilot entries in this changelog no longer apply. Batch tasks with
taskType: "COPILOT"
are rejected with
MODEL_UNAVAILABLE
and are not created. To track Bing’s AI answers, use
Bing AI Mode
. See
unsupported providers
.
Navigate to changelog: 6th November 2026
Google Search: redirect links resolve again
On 6 November Google closed the method cloro used to resolve organic result redirects (
/goto?url=…
). We deployed a new method a few hours later, and the
Google Search endpoint
returns resolved links again.
If a response from those hours has an empty
link
, pass its
redirectLink
to
POST /v1/monitor/google/goto
to get the destination. It costs 1 credit, charged only when the destination comes back. The endpoint also resolves
/goto
links that you did not get from cloro.
Navigate to changelog: 2nd November 2026
ChatGPT: answers kept when ChatGPT does not search or cite
The
ChatGPT endpoint
forces web search, and it used to fail any attempt that showed no sign of search. Two things changed:
An answer where ChatGPT searched but cited nothing now succeeds, with empty sources and citationPills .
If ChatGPT does not search on attempts 1 and 2, cloro keeps the answer from the third attempt on, searched or not.
Successful responses can now have empty
sources
,
citationPills
, and
searchQueries
even without
disableWebSearch
. Retrying these prompts rarely changes the result. See
Why are
sources
and
citationPills
empty when web search is forced?
Navigate to changelog: 30th October 2026
Google Search: resolving a redirect link now costs 1 credit
POST /v1/monitor/google/goto
now costs 1 credit per request, with no sync surcharge. You pay only when the call returns the destination, so a failed call costs nothing.
Demand for this endpoint has grown very fast, because the industry has not been able to solve these links, and each link takes significant processing to resolve. We have to charge for it to sustain this demand.
Navigate to changelog: 30th October 2026
Google Search: add a fallback for unresolved links
Google now puts most organic result redirects (
/goto?url=…
) behind a verification page, so resolving them can take longer. This affects every SERP provider we track. cloro used to fail a page when a link stayed unresolved, so from 01:40 to 06:50 UTC on 30 October about 94% of requests to the
Google Search endpoint
failed. Failed requests were not charged.
The endpoint now returns the page, and a link that is not resolved yet has an empty
link
:
{
"position"
:
3
,
"title"
:
"Understand the basics of Roth IRAs"
,
"link"
:
""
,
"redirectLink"
:
"https://www.google.com/goto?url=CAESaAHrOzAVev0pi\_wnTxm1..."
}
We still try to resolve each link in the response, but you should always have a fallback. When
link
is empty, pass
redirectLink
to
POST /v1/monitor/google/goto
to get the destination. If the call returns
502
, retry it later.
Navigate to changelog: 29th October 2026
ChatGPT: product rationale and reviews
Set
include.productDetails: true
with
include.shopping: true
on the
ChatGPT endpoint
to get the “What to know” rationale and the “What people are saying” reviews that ChatGPT shows when you open a product. They come back as
productDetails
on each product in
shoppingCards
and
inlineProducts
, and on each
entity listing
in
entities[n].product
. This adds +2 credits to the request, on top of the +2 credits for
include.shopping
. See
product details
.
ChatGPT: all offers on shopping products
offers
on shopping-card and inline products now has every offer ChatGPT shows under “Compare offers”. Before, it had only the first one. Answers from ChatGPT’s mobile interface now have
offers
and
url
too, as desktop answers already did.
ChatGPT: gpt-6 product cards are in
inlineProducts
Each gpt-6 product card is now an inline product. A card with a listing has it in
inlineProducts
, not in
entities[n].product
. An entity keeps its listing only when it is outside a card. A card with no listing is an inline product too: its
product
comes from the card (title, link, price text, badge, subtitle, description, “Label: value” lines such as “Best for: …” in
attributes
, and image), with no
id
and no
offers
. Each card also has
citationPillIds
: the
citationPillId
of each citation chip in the card, so you can find in
result.citationPills
which sources back each product.
Navigate to changelog: 27th October 2026
Gemini: responses about 60% faster
The
Gemini endpoint
now returns a response in about 14 seconds on average, against about 37 seconds before.
include.html
adds about 2 seconds.
Navigate to changelog: 27th October 2026
Google Search: degraded results are back
Since 26 October, about 20:10 UTC, Google again gives most sessions a
degraded SERP
, across every provider we track. The organic results then show pages that are similar to the query but are not the sites that usually rank for it. Requests to the
Google Search endpoint
can return this degraded page until it is fixed.
We’re working on it, and we’ll post here when full results are back.
Navigate to changelog: 21st October 2026
Async: the idempotency
409
returns the existing task ID
When you reuse an
idempotencyKey
, the
409 RESOURCE_ALREADY_EXISTS
from
POST /v1/async/task
now has
error.details.taskId
: the ID of the task that already holds the key. If your submit request timed out before you received the
taskId
, resubmit with the same key and poll the returned ID. The
batch endpoint
returns it too, except for a key repeated inside the same batch.
Google: the AI Overview is expanded in the HTML
When a
Google Search
request sends
include.aioverview
and
include.html
, the stored HTML now shows the AI Overview expanded, as it looks after a click on “Show more”. Before, it was collapsed. The cost does not change.
Navigate to changelog: 19th October 2026
Google: new
recipes
field for the Recipes carousel
The
Google Search endpoint
now returns the cards of Google’s Recipes carousel as
result.recipes
, on desktop and mobile, with no flag and at no extra cost. Each card has
position
,
title
and
link
plus, when Google shows them,
source
,
rating
,
reviews
,
totalTime
,
ingredients
and
thumbnail
. The field is omitted when the SERP has no Recipes carousel.
On desktop these cards were returned in
result.peopleAreSaying
by mistake.
peopleAreSaying
now has only the cards of the “What people are saying” module.
Navigate to changelog: 16th October 2026
ChatGPT: entities now have their product listing
When ChatGPT attaches a product listing to an entity, the
ChatGPT endpoint
now returns it as
entities[n].product
: the price, the URL and the merchant offers, in the same structure as a shopping card product. It needs
include.shopping: true
, at no extra cost over that add-on.
Navigate to changelog: 14th October 2026
ChatGPT: multi-turn conversations
The
ChatGPT endpoint
now accepts
prompt
as an array of up to 5 prompts. ChatGPT answers them in order, in the same conversation, and
result
is an array with one response for each prompt. Use it to measure what ChatGPT says, cites and advertises after the first answer.
{
"prompt"
: [
"What are the best running shoes for flat feet?"
,
"Which of these costs less than $100?"
],
"country"
:
"US"
}
Each prompt after the first costs +1 credit. A
prompt
string works as before and returns one object.
ChatGPT currently shows ads only in answers to a follow-up prompt, not in the answer to the first prompt of a conversation. To get
result.ads
, send two or more prompts with
include.ads: true
and
legacy: true
, and read the ads from the second response or a later one. Follow-ups on the default mobile-web interface return ads far less often.
Navigate to changelog: 14th October 2026
Google Search: fewer degraded results
Google gives some sessions a degraded SERP: fewer pages, and fewer of the sites that usually rank for the keyword. A site can then drop out of the results even though its rank did not change. The
Google Search endpoint
now sends a request only through a session that a test confirms gets Google’s full results, so ranks are more stable from one request to the next.
Navigate to changelog: 12th October 2026
Google:
localResults
on mobile
The
Google Search endpoint
now returns the
local pack
for
device: "mobile"
in all the layouts Google shows. Before, most mobile responses had no
localResults
. On mobile,
phone
is the number from the card’s Call button, such as
"+17027773777"
.
On desktop and mobile, a price in a currency that has no symbol in the text, such as
"20–40 zł"
or
"100-200 kr."
, is now in
price
. Before, it was in
type
and the category was lost.
Navigate to changelog: 10th October 2026
ChatGPT: the follow-up form is now in the response
ChatGPT now ends many answers with a form that refines the prompt, such as “What’s your budget?” with a list of ranges. The
ChatGPT endpoint
returns it as
result.followUpForm
, with no flag and at no extra cost: the questions, their options, the one ChatGPT selects by default, and the follow-up prompt it would send.
followUpForm
is
null
when the answer has no form. The form text no longer shows in
text
and
markdown
.
Navigate to changelog: 9th October 2026
AI Mode: one ID links shopping cards, inline products and product results
Google AI Mode
shopping cards, inline products and product results now have a
productDocid
, which is Google’s ID for the product. Entries for the same product in one response have the same value, so you can
join the three blocks
without a match on title, store or price.
Google can change the value between responses, so do not use it to identify a product over time. The field is absent when the link from Google has no ID.
Navigate to changelog: 8th October 2026
AI Mode: product tokens replace
include.expandProducts
Google AI Mode
shopping cards and inline products that link to Google’s product viewer now have a
productResultToken
, with no flag and at no extra cost. Send it to
POST /v1/monitor/google/product
within five minutes to get the product’s
merchant offers
. Each product request costs 1 credit (3 sync), so you pay only for the products you need.
include.expandProducts
is deprecated. From 23rd October 2026 the API ignores it: a request that sends it still succeeds, but returns no
result.productResults
. Use the token.
Navigate to changelog: 8th October 2026
Google Search: get the product panel behind a shopping card
A Google shopping card opens a product panel, so its
productLink
is usually empty. Each desktop first-page card now has a
productResultToken
. Send it to the new
POST /v1/monitor/google/product
endpoint to get the panel as
productResults
: merchant offers with links and prices, specifications, reviews, videos, variants and related products.
{
"productResultToken"
:
"nuzK-QgAV8BqrZdwQJYZopzdoFhpvKr-N2_YDZevwzg"
}
A token expires five minutes after the search. The search costs no more than before. The product request costs 1 credit (3 sync), and only when it succeeds. It is also available async as
taskType: "GOOGLE_PRODUCT"
. Set a high
priority
on an async product request, so that the token does not expire in the queue.
Navigate to changelog: 29th September 2026
Send requests from the dashboard Playground
The
Playground
now sends requests to every monitoring endpoint, with no API key. Pick an endpoint, fill in its parameters, and see the response in three tabs:
JSON : the full response.
HTML : the page that the endpoint returned, rendered.
Code : the same request in cURL, Python and Node.js, ready to copy.
Requests use your organization’s credits and show in
Logs
. The page URL keeps the parameters, so you can share a request with your team.
Navigate to changelog: 24th September 2026
ChatGPT: Czech Republic (
CZ
) is supported again
The
ChatGPT endpoint
accepts
country: "CZ"
again. We removed it on 30th June 2026 because ChatGPT showed a login wall to anonymous users from Czech IPs. ChatGPT no longer shows that wall, and answers for
CZ
are localized to the Czech Republic. The
/v1/countries
endpoint lists
CZ
under
model=chatgpt
again.
Navigate to changelog: 22nd September 2026
Gemini product cards now ship as
shoppingCards
Gemini shows product cards in shopping answers. Before, the response had no structured data for them. Now each card comes back as an object, with the same field names as AI Mode’s
shoppingCards
and more fields that only Gemini supplies:
{
"position"
:
1
,
"title"
:
"ASICS Men's Novablast 5"
,
"productLink"
:
"https://google.com/search?q=ASICS\+Men%27s\+Novablast\+5&prds=...&ibp=oshop"
,
"productId"
:
"11413473167792520357"
,
"price"
: {
"value"
:
129.95
,
"currency"
:
"$"
,
"raw"
:
"$129.95"
},
"oldPrice"
: {
"value"
:
149.99
,
"currency"
:
"$"
,
"raw"
:
"$149.99"
},
"store"
:
"Marathon Sports"
,
"storeLink"
:
"https://www.marathonsports.com/products/asics-mens-novablast-5"
,
"storeCount"
:
2
,
"shipping"
:
"Free"
,
"deliveryTime"
:
"Up to 5 days"
,
"rating"
:
4.8
,
"reviews"
:
11558
,
"brand"
:
"ASICS"
,
"thumbnail"
:
"https://encrypted-tbn1.gstatic.com/shopping?q=tbn:example"
,
"catalogId"
:
"15000182008314641842"
,
"gpcId"
:
"1340101364570861215"
}
Cards also carry
description
and
images
. See the
shopping cards schema
. Shopping cards are intent-detected, so no flag is required and there is no extra cost.
shoppingCards
is absent when the answer has no product cards.
markdown
,
sources
and
places
do not change.
Navigate to changelog: 21st September 2026
AI Mode product results report each market’s own currency
result.productResults[].stores[].price.currency
now returns the correct ISO code for Brazil, Mexico and the UAE. Before, Brazilian offers (
"R$ 5.597,99"
) were mislabelled as
"USD"
, and Mexican (
"MXN 19,499.00"
) and Emirati (
"3,399.00 د.إ"
) offers lost their price entirely. They now return
"BRL"
,
"MXN"
and
"AED"
.
A bare
$
now follows the request
country
, so
country: "CA"
returns
"CAD"
and
country: "AU"
returns
"AUD"
. Other countries keep
"USD"
. An explicit currency token always takes priority over the country.
Google Search product results
also return
"BRL"
for
R$
prices. See
currency codes
.
AI Mode ads carry the creative image as
imageUrl
AI Mode ad cards now return the creative as a public
https://tpc.googlesyndication.com/simgad/...
URL on
imageUrl
. Roughly a third of AI Mode ads carry a creative; text ads without one omit the field, so treat it as optional.
{
"title"
:
"Bambu Lab X2D"
,
"url"
:
"https://www.google.com/aclk?sa=L&ai=abc"
,
"position"
:
1
,
"imageUrl"
:
"https://tpc.googlesyndication.com/simgad/1234567890123456789"
}
The change is additive; no request parameter is needed. See the
AI Mode ads schema
.
Navigate to changelog: 18th September 2026
Place cards carry
phone
and
website
, and
placeId
is now a Google ID
Place cards render Call, Directions and Website buttons.
places[]
and
map[]
now read all three, on AI Mode and AI Overview alike:
{
"title"
:
"The Dagny Boston"
,
"placeId"
:
"/g/1tglgvbk"
,
"address"
:
"89 Broad St, Boston, MA 02110"
,
"phone"
:
"+18776865313"
,
"website"
:
"https://www.thedagnyboston.com/"
}
address
is the full postal address from the Directions link. It used to be the short line printed on the card — a street line, a “city, region” line, or nothing at all.
placeId
was a hash of the title and address. It is now the place’s Google Knowledge Graph ID, the same id its
map
entry carries, so the two join on it. Two things change for you: the value differs from what you stored before, and it is omitted on a card Google did not geo-tag.
phone
and
website
are omitted when the card renders no Call or Website button.
See the
places
and
map
schemas.
Navigate to changelog: 18th September 2026
Google AI Overview ships place cards as
places
and
map
Local answers show place cards. Their Directions and Website buttons used to reach you in
sources
; the cards now come back as
result.aioverview.places
and
result.aioverview.map
, with the same structure as in AI Mode.
On both surfaces:
places[].link is the place’s Google viewer URL, the same URL as its entry in sources .
map[].link is a Google Maps URL that opens the place, for example https://maps.google.com/?cid=11340480041048787817 .
See the
AI Overview schema
.
Navigate to changelog: 16th September 2026
Perplexity is back to normal success rates
The
Perplexity endpoint
was degraded over the past weeks, and we’re happy to report that it is now back to normal.
Perplexity has been throwing log-in walls on nearly all requests. We worked around the issue and the endpoint is now fully functional again.
Nothing about the endpoint changed. Same price, same request parameters, same response shape.
Per-incident timings are on the
status page
, and you can subscribe there to get the next one by email.
Navigate to changelog: 14th September 2026
Google shopping cards ship the discount badge as
promoLabel
A discounted card renders a badge above the product name. That badge was reaching you as
title
, in the SERP language —
LOW PRICE
,
OFERTA
,
PROMOÇÃO
— and the name was dropped.
title
now always carries the name, and the badge comes back beside it:
{
"title"
:
"Dove Jabón Barra Original 135 g"
,
"promoLabel"
:
"OFERTA"
,
"price"
: {
"value"
:
8.79
,
"currency"
:
"$"
,
"raw"
:
"$8.79"
},
"oldPrice"
: {
"value"
:
11
,
"currency"
:
"$"
,
"raw"
:
"$11"
}
}
Omitted on a card without a badge. See the
shopping cards schema
.
Navigate to changelog: 9th September 2026
Gemini place cards now ship as
places
Lodging and local answers carry a card carousel. Its photos used to land in
sources
as image URLs while the place itself was dropped; the cards now come back as their own array, and
sources
holds only real citations.
{
"index"
:
0
,
"title"
:
"Pullman Quay Grand Sydney Harbour"
,
"link"
:
"https://www.google.com/travel/search?q=Pullman\+Quay\+Grand\+Sydney\+Harbour"
,
"rating"
:
4.4
,
"reviews"
:
1307
,
"type"
:
"5-star hotel"
,
"price"
:
"$565"
}
Intent-detected, so no flag is required. See the
places schema
.
Navigate to changelog: 8th September 2026
Every redirect link now ships beside its destination as
redirectLink
Google serves many result links as an encrypted
/goto?url=<token>
redirect. cloro resolves them, and now returns the redirect too, so you no longer choose between the two:
{
"position"
:
1
,
"title"
:
"24 best SEO tools I'm using in 2026 (free + paid)"
,
"link"
:
"https://www.marketermilk.com/blog/best-seo-tools"
,
"redirectLink"
:
"https://www.google.com/goto?url=CAESaAHrOzAVev0pi\_wnTxm1qpiZTmMyAllabLr1or10ZCfvJKPk"
,
"page"
:
1
}
It appears on any link field Google served as a redirect, across
Google Search
,
Google News
and
AI Mode
, and is the absolute form
POST /v1/monitor/google/goto
takes.
Every field now holds a destination, including the ones that still returned a raw token until early September:
peopleAlsoAsk[].sources[].url
,
newsResults[].link
,
productResults[].stores[].link
, and the links inside
markdown
.
The key is omitted when Google served a plain link, so its absence means there was no redirect. Sponsored results are the exception: an ad’s
/aclk
URL is a tracking link, not a redirect, and following it registers a click against the advertiser — so ads carry no
redirectLink
on Google Search.
include.googleGoto
is now ignored.
The flag returned raw
/goto
links
instead of
destinations. Requests still sending it get a normal response, with the destination in
link
and the redirect in
redirectLink
— more than the flag gave you, so nothing needs to change. One exception: links inside
markdown
and
text
are resolved destinations and no longer carry the raw token. Read
redirectLink
from the structured fields instead.
POST /v1/monitor/google/goto
is unaffected.
A redirect we cannot resolve now omits the link field rather than returning a token you cannot fetch.
redirectLink
is still there, so a missing
link
or
url
means “unresolved”. Pass it to
POST /v1/monitor/google/goto
to get the destination.
Navigate to changelog: 2nd September 2026
ChatGPT: opt out of forced web search with
disableWebSearch
The
ChatGPT endpoint
forces ChatGPT’s web search on every request. Set
disableWebSearch: true
and cloro returns the answer ChatGPT gives on its own, searching only when it decides to. Most prompts then come back without search, so
result.sources
,
result.citationPills
, and
result.searchQueries
are often empty, and the answer text can differ from the search-forced one. Use it to see what ChatGPT says without search, and keep the default to measure citations. The cost is unchanged, and requests that do not send the flag behave exactly as before.
Navigate to changelog: 31st August 2026
Google and AI Mode: keep the raw
/goto
links, and a new endpoint to resolve one
cloro resolves Google’s encrypted
/goto?url=<token>
redirect links for you, so result links are real destinations. If you run your own resolution, set
include.googleGoto: true
on a
Google Search
or
AI Mode
request and the links come back raw instead, as absolute
https://www.google.com/goto?url=<token>
URLs:
{
"position"
:
1
,
"title"
:
"24 best SEO tools I'm using in 2026 (free + paid)"
,
"link"
:
,
"page"
:
1
}
POST /v1/monitor/google/goto
resolves one of those links to its destination. It charges no credits and uses no concurrency slot.
The flag is off by default, so requests that do not send it behave exactly as before. See
Google redirect links
.
Superseded on 8th September 2026:
include.googleGoto
is ignored. Every redirect now ships as
redirectLink
beside its destination, so you get both without the flag.
Navigate to changelog: 30th August 2026
Google Search: sponsored-ad coverage roughly doubled
cloro now targets geography on
Google Search
requests in a way that no longer suppresses Google’s ad auction, so sponsored results that were previously withheld come through. In a paired A/B test on commercial-intent US queries, ads roughly double the coverage.
Nothing changes in your integration. The
country
request parameter (and its
gl
alias) targets the same geography as before,
hl
still controls the interface language, and
sponsored ads
keep arriving in
result.ads
with the same schema. You should simply see ads on more commercial queries.
Navigate to changelog: 28th August 2026
ChatGPT: request the legacy interface with
legacy
The
ChatGPT endpoint
accepts a
legacy
boolean. Set it to
true
and cloro asks ChatGPT for its legacy desktop interface instead of the mobile-web one that now serves almost all traffic. That interface streams the answer as an event stream, so
result.model
,
result.searchQueries
, and
result.mapSearchQueries
come back filled in rather than empty. Unset, requests behave exactly as before.
This is a temporary workaround, not a supported switch. OpenAI decides which interface it serves and is rolling out mobile-web as the default, so
legacy: true
is best-effort and can stop working without notice. Keep your parsing working for both shapes: even with the flag set, treat
result.model
,
result.searchQueries
, and
result.mapSearchQueries
as possibly empty, and handle
rawResponse
in both the event-stream and HTML-fragment shapes.
Navigate to changelog: 27th August 2026
/goto
tracking links: your results keep real URLs
Google now serves many result links as relative
/goto?url=<token>
redirects instead of the destination URL. The token is encrypted on Google’s side, so the raw HTML gives no usable link.
cloro resolves them before the response leaves us. The
/goto
links stay in the HTML, but the parsed JSON holds the real destination:
{
"organicResults"
: [
{
"position"
:
1
,
"title"
:
"Best laptops for programmers"
,
"link"
:
"https://example.com/best-laptops-2026"
,
"displayedLink"
:
"example.com › best-laptops"
}
]
}
For the few links we cannot resolve this way, we follow the
/goto
link to its destination. This adds a small delay to those requests only.
Nothing changes in your integration.
link
and the video and local-pack link fields stay absolute
https://
URLs, as before. If a Google organic link cannot be resolved at all, we retry the request instead of giving you a
/goto
value.
Navigate to changelog: 26th August 2026
ChatGPT: the mobile-web format now serves 98% of responses
OpenAI completed the rollout of the mobile-web response format on
25 August 2026
. It served approximately 10% of successful ChatGPT responses through early August, approximately 50% between 18 and 24 August, and
98%
from 25 August.
Nothing changed in the cloro API. The effects are the ones in the 21st July 2026 entry below:
result.model
,
result.searchQueries
, and
result.mapSearchQueries
come back empty,
result.sources
holds fewer entries, and
rawResponse
uses the HTML-fragment shape. The answer
text
/
markdown
and inline
citationPills
are unaffected.
An empty field does not mean the field is gone. Providers have restored this kind of metadata later, so keep your parsing for
result.model
,
result.searchQueries
, and
result.mapSearchQueries
in place.
OpenAI selects the format per request on their side. cloro replicates a real user and does not force a particular interface, so we cannot select the event-stream format for you.
Navigate to changelog: 25th August 2026
Google family: interface language with
hl
Google Search
,
Google News
, and
AI Mode
take a new optional
hl
parameter for the interface language, independent of the country:
{
"prompt"
:
"beste laptops für programmierer"
,
"country"
:
"US"
,
"hl"
:
"de"
}
Before this, the language came from the country, so a German interface meant asking for Germany, and multilingual countries such as
CH
,
BE
, and
CA
picked one dominant language. AI Mode had no override at all.
The same three endpoints also accept
gl
as an alternative name for
country
, matching Google’s own parameter. Both work, and
gl
takes either case; sending
country
and
gl
with different values is a
400
. Nothing changes for existing requests.
Perplexity pricing increases to 4 credits per request
Perplexity continues to restrict access. To hold the same success rate and source coverage, our cost to serve this endpoint increased by 70%. The
Perplexity endpoint
now costs
4 credits per request
, up from 3. That is an increase of 33% against a cost increase of 70%. All other provider prices, addons, and the sync surcharge do not change. See the full
providers pricing table
for context.
Navigate to changelog: 17th August 2026
Scheduled maintenance: 19 August 2026, 18:00 UTC
We will do routine infrastructure upgrades on
Wednesday 19 August 2026 from 18:00 UTC
. They will take approximately 1 to 2 hours.
Scraping and the API stay fully operational for the full window. Dashboard analytics and log views can be unavailable for short periods, which leaves a short gap in the dashboard data for that window. No action is necessary.
Navigate to changelog: 14th August 2026
Request latency is now returned in responses
Every API response carries an
X-Latency-Ms
header with the milliseconds the API spent on that request. It excludes network transit.
Async task processing time is reported separately as
task.latencyMs
, measured from first pickup to the final outcome and excluding the initial queue wait. It is returned by
GET /v1/async/task/{taskId}
, in the
webhook payload
, and in create responses, where it is
null
until the task finishes.
Both additions are additive — no existing field or header changed.
Navigate to changelog: 11th August 2026
Incident:
result.html
unavailable between 12:00 and 12:47 UTC
A DNS misconfiguration meant the URLs returned in
result.html
did not resolve between
12:00 and 12:47 UTC on 11 August 2026
. Requests completed and were charged as normal, and every other
result
field was unaffected. DNS has been corrected.
The HTML from that window cannot be recovered. Contact support with the number of affected credits and we will reimburse them.
Navigate to changelog: 10th August 2026
Google Search and Google News:
device
now accepts
ios
and
android
The
device
parameter on the
Google Search
and
Google News
endpoints takes two new values:
ios
, which runs the search as Safari on an iPhone, and
android
, which runs it as Chrome on Android. Both return the mobile SERP layout, so the response shape is unchanged and there is no extra credit cost.
mobile
still works and is now an alias for
android
, so existing requests behave exactly as before. Use
ios
when you need to see what Google serves iPhone users specifically.
Merchant offers on Google Search and AI Mode:
result.productResults
Both Google surfaces can now return the merchant offers behind a product — direct merchant URL, price, installment terms, and stock, delivery and returns badges — as a new
productResults
array, in the same shape on both.
On
Google Search
it appears whenever Google’s right-rail product panel does, with no flag and no extra credits;
shoppingCards
is unchanged. On
Google AI Mode
it is opt-in via
include.expandProducts
and costs
+1 credit per product cluster returned
, capped at +6.
Navigate to changelog: 7th August 2026
MCP server
cloro now ships a hosted
MCP server
that exposes the monitor endpoints as agent tools for any MCP client — Claude, Claude Code, Cursor, or your own agent. Connect over Streamable HTTP at
https://mcp.cloro.dev/YOUR\_CLORO\_API\_KEY/mcp
and ask your agent how AI assistants answer a prompt, whether a brand appears in Google’s AI Overview, or what Google News is saying — no HTTP calls to hand-write. The server forwards your own API key to the API, so billing, rate limits, and concurrency work exactly as they do for direct requests.
See the
MCP server guide
for client configuration and the full tool list.
Navigate to changelog: 30th July 2026
New endpoint: read your credit balance with
GET /v1/credits
The new
GET /v1/credits
endpoint returns your current balance (
remaining
), credits per cycle (
perCycle
), and the cycle reset date (
cycleResetsAt
). Async-only workloads can now read their balance and reset date programmatically — for low-balance alerts or a submission breaker — without making a billable sync
/v1/monitor/*
call just to inspect the
X-Credits-Remaining
header.
Navigate to changelog: 27th July 2026
Google AI Mode: 33 newly-supported countries
The
Google AI Mode endpoint
now accepts 33 additional countries and territories that Google has rolled out since the last update, bringing its supported list in line with the
official AI Mode availability list
. Every country accepted by AI Mode is also accepted by the
Google Search endpoint
, so you can now freely use the same
country
value across both.
The additions are:
BL
,
BV
,
BY
,
EH
,
ER
,
ET
,
GF
,
GP
,
GU
,
GW
,
IS
,
KM
,
KZ
,
LK
,
LR
,
MC
,
MF
,
MO
,
MQ
,
MR
,
NC
,
PF
,
PM
,
RE
,
SY
,
SZ
,
TF
,
TZ
,
VA
,
WF
,
XK
,
YE
,
YT
. The
/v1/countries
endpoint now lists them under
model=aimode
.
Navigate to changelog: 22nd July 2026
Google AI Mode and AI Overview: France (
FR
) is now supported
You can now send
country: "FR"
to the
Google AI Mode endpoint
and use
include.aioverview
on the
Google endpoint
for
FR
requests. Previously these were rejected as unsupported regions; regular Google Search already accepted
FR
. The
/v1/countries
endpoint now lists
FR
under
model=aimode
and
model=aioverview
.
Navigate to changelog: 21st July 2026
ChatGPT: new mobile-web response format
OpenAI now serves a share of logged-out ChatGPT traffic through a new mobile-web format that streams the answer as HTML fragments rather than the usual event stream. On responses served this way:
rawResponse (when you opt in) has a different, HTML-fragment shape alongside the existing event-stream shape.
result.model , result.searchQueries , and result.mapSearchQueries come back empty — the mobile format carries none of the stream metadata these are parsed from.
result.sources is shorter ; the format exposes fewer full source entries.
The answer text / markdown and inline citationPills are unaffected .
Which format OpenAI serves is decided on their side, per request. cloro replicates a real user, so it does not — and by design should not — force a particular interface; a share of responses will therefore arrive in the mobile-web format.
Fields that were previously always populated can now be empty on mobile-web responses. Treat
result.model
,
result.searchQueries
, and
result.mapSearchQueries
as possibly empty, and expect fewer entries in
result.sources
. If you consume
rawResponse
, handle the HTML-fragment shape in addition to the event-stream shape.
Navigate to changelog: 20th July 2026
ChatGPT:
visibility
,
adsResponseIndex
, and
rendered
on ads
Each ad in
result.ads
(set
include.ads: true
) now carries three fields that separate what OpenAI
served
from what the user actually
saw
. OpenAI serves candidate ads but the interface renders at most one; previously
result.ads
listed the served candidates with no way to tell which was shown.
rendered
is
true
only for the ad displayed as a visible card, so you can count real impressions instead of the array length.
visibility
exposes OpenAI’s serving status (
allowed
, or
hidden
when an ad was filtered), and
adsResponseIndex
gives the ad’s served slot position. The fields ride the existing
include.ads
flag, so there is no new parameter and no extra credit cost. See the
ChatGPT ads schema
.
Navigate to changelog: 18th July 2026
New
DELETE /v1/async/queue
endpoint to clear your async queue
You can now clear your organization’s pending async queue in a single call with the new
Clear queue endpoint
. It deletes every task still in the
QUEUED
state and returns the number removed (
{ "success": true, "cleared": <count> }
), giving you a self-serve way to drain a backlog instead of waiting for it to process. Tasks already
PROCESSING
are in-flight and left running, and
COMPLETED
/
FAILED
tasks are untouched. Queued tasks aren’t charged until they’re processed, so clearing them has no effect on your credit balance, and the call is idempotent — an empty queue simply returns
cleared: 0
. Check how many tasks are queued first with the
async status endpoint
.
Navigate to changelog: 18th July 2026
Perplexity:
search_model_queries
is now an array of strings
Around
15th July 2026, Perplexity changed the structure of its underlying response object
— it moved the model’s search queries into a new location and stopped sending the per-query engine and result-limit metadata. During the interim, that upstream change had left
result.search_model_queries
empty on most responses.
This update adapts to the new structure and restores the field. The
Perplexity endpoint
now returns
result.search_model_queries
as a plain array of query strings, matching how ChatGPT exposes the same fan-out concept. Each entry was previously an object with
query
,
engine
, and
limit
fields;
engine
was always
"web"
and
limit
was always
8
, stale constants that no longer corresponded to anything Perplexity sends. We’ve dropped them and kept only the query text.
This is a breaking change to the shape of
search_model_queries
entries. If you read
.query
off each entry, switch to reading the string directly (
result.search_model_queries[i]
instead of
result.search_model_queries[i].query
); if you depended on
.engine
or
.limit
, those values were fixed constants and are no longer emitted.
Navigate to changelog: 10th July 2026
ChatGPT: new
mapSearchQueries
field
The
ChatGPT endpoint
now exposes a
mapSearchQueries
array on
result
when you set
include.searchQueries: true
. When ChatGPT renders a map block for a local-intent prompt (e.g. “best coffee shops in San Francisco”, “plumbers near me”), it fires a separate query at its maps tool to gather the businesses behind
result.map
. That query previously lived only in the raw stream. It now has a structured home, distinct from
result.searchQueries
(the web-search fan-out), which the two can diverge on for the same prompt. The field rides the existing
include.searchQueries
flag, so there is no new parameter and no extra credit cost. It is additive and omitted when no map appears in the response. See the
ChatGPT response schema
.
Navigate to changelog: 9th July 2026
Google: new
localResults
(local pack / “3-pack”) field
The
Google Search endpoint
now exposes a
localResults
array on
result
for local-intent queries (e.g. “best pizza in new york city”, “plumbers in houston”). Each entry is a place from Google’s map-backed local pack, carrying
position
and
title
plus, when Google renders them,
placeId
,
rating
,
reviews
,
price
,
type
,
yearsInBusiness
,
address
,
phone
,
hours
, a
description
snippet, and a
links
object with the place’s
website
and
directions
URLs (typical on business/service packs). When the pack shows a “More places” / “More businesses” control,
result.localResultsMoreLink
carries the URL of Google’s expanded Local Finder for the query. The field is additive and omitted when no local pack appears on the SERP; extraction is desktop only. See the
Local Pack schema
.
Navigate to changelog: 8th July 2026
Google AI Overview: new
relatedLinks
field
Google AI Overview responses now expose a
relatedLinks
array on
result.aioverview
. These are a citation chip’s “View related links” flyout items — URLs Google groups under a pill that are
not
in the
sources
rail (for example a Google Shopping comparison link). Previously these rendered inline in
markdown
with no structured home, which made
markdown
appear to carry more URLs than
sources
/
citationPills
. Each related link now shares the
citationPillId
of its chip (so you can group it with that chip’s citation pills) and carries
label
,
url
, and
domain
. The field is additive and omitted when no chip carries related links. See the
AI Overview schema
.
Navigate to changelog: 7th July 2026
Copilot fully back, with Brazil (BR) and Italy (IT) re-enabled
The
/v1/monitor/copilot
endpoint is fully operational again. As part of the same rollout,
country: "BR"
and
country: "IT"
are now accepted — requests previously rejected with an unsupported-country error will route through normally.
Navigate to changelog: 2nd July 2026
OpenClaw integration
cloro now ships a plugin for
OpenClaw
that exposes the monitor endpoints as native agent tools —
cloro_chatgpt
,
cloro_perplexity
,
cloro_gemini
,
cloro_copilot
,
cloro_aimode
,
cloro_google
, and
cloro_google_news
. Ask your OpenClaw agent how AI assistants answer a prompt, whether a brand appears in Google’s AI Overview, or what Google News says — no HTTP calls to hand-write.
See the
OpenClaw integration guide
for installation and configuration, including an MCP-server alternative.
Navigate to changelog: 30th June 2026
Czech Republic (CZ) targeting disabled in ChatGPT
We’ve removed
country: "CZ"
support for the ChatGPT endpoint. ChatGPT no longer allows anonymous access from Czech IPs — every request is met with a login wall, which scrapers cannot pass through. Requests with
country: "CZ"
to
/v1/chatgpt
will now return an unsupported-country error instead of silently failing.
If you target CZ on other scrapers, double-check the results you’re getting back. Some providers fall back to proxies in neighboring countries when CZ-resident proxies are unavailable, which can skew localized answers.
Navigate to changelog: 30th June 2026
Google Search and AI Mode now pin to the requested country by default
Requests to
/v1/monitor/google
and
/v1/monitor/aimode
that pass
country
but omit both
location
and
uule
now auto-default
location
to the country’s canonical name (e.g.
"Portugal"
for
country: "PT"
). Previously, with no UULE set, Google geolocated by the proxy’s exit IP — so identical sequential requests could swing between different result locations as each scrape leased a different proxy session.
No request changes are needed. Existing calls will see more stable, country-pinned results. Defaulting is country-level only — cloro never auto-picks a city. If you need city- or region-level precision, keep passing
location
(Google canonical name) or
uule
(pre-encoded UULE) explicitly. Unknown country codes leave UULE unset, preserving prior behavior.
{
"query"
:
"best restaurants"
,
"country"
:
"PT"
}
Navigate to changelog: 29th June 2026
Google Knowledge Graph panel now extracted
The Google Search endpoint now returns a
knowledgeGraph
object when a Knowledge Graph panel appears on the SERP. The panel covers people, places, organizations, films, local businesses, and other well-known entities — surfacing structured data like entity type, description, attributes, profiles, ratings, and panel-type-specific fields (streaming options, admission prices, social posts, and more).
No request changes needed — the field is included automatically when a panel is present, and omitted from
result
when it is not.
{
"result"
: {
"knowledgeGraph"
: {
"title"
:
"Eminem"
,
"type"
:
"American rapper"
,
"description"
:
"Marshall Bruce Mathers III, known professionally as Eminem..."
,
"website"
:
,
"attributes"
: [
{
"key"
:
"Born"
,
"value"
:
"October 17, 1972"
},
{
"key"
:
"Genre"
,
"value"
:
"Hip hop"
}
]
}
}
}
See the
Knowledge Graph reference
for the full schema.
Navigate to changelog: 24th June 2026
State-level geo-targeting for AI endpoints
You can now target a specific US state when querying ChatGPT, Copilot, Perplexity, Gemini, or Grok. Pass the USPS two-letter code as the
state
parameter alongside
country: "US"
and cloro routes the request through a proxy in that state — giving you localized results at the sub-country level.
{
"prompt"
:
"best electricians near me"
,
"country"
:
"US"
,
"state"
:
"TX"
}
State targeting costs
+2 credits
on top of the base cost and any other add-ons. Use
GET /v1/states?country=US
to retrieve the full list of supported state codes. Google Search and AI Mode are not affected — those endpoints continue to use
location
/
uule
for sub-country targeting.
Navigate to changelog: 22nd June 2026
Rate limits are now per endpoint
The API rate limit of 1,000 requests per second now applies independently to each endpoint (
/v1/*
) per API key. Previously, all endpoints shared a single bucket, meaning a burst on one endpoint could exhaust the limit for others. Each endpoint now has its own independent bucket, so high-traffic calls to one endpoint no longer affect the rate limit for any other.
No action required — existing integrations are unaffected. The
X-RateLimit-Limit
and
X-RateLimit-Remaining
response headers continue to reflect the limit and remaining requests for the specific endpoint called.
Navigate to changelog: 17th June 2026
Ask the docs — AI assistant now available
You can now ask questions about cloro directly from the docs. The assistant is trained on all published documentation and answers in natural language, with citations linking back to the relevant pages.
Look for the
Ask Assistant
button in the navigation bar or the search bar. It works across all pages — API reference, guides, and the changelog.
Navigate to changelog: 1st June 2026
Webhook signing — verify deliveries came from cloro
Enable webhook signing from the
dashboard
and every async-task delivery will carry an
X-Cloro-Signature
header (HMAC-SHA256 over
<timestamp>.<raw_body>
) plus
X-Cloro-Timestamp
and
X-Cloro-Webhook-Id
. Your endpoint can confirm the payload’s origin and integrity, and reject anything signed more than a few minutes ago to protect against replay.
The new
Webhooks guide
is the single home for everything that happens at your endpoint — payload shape, retry behavior, signature verification (with copy-pasteable Node.js, Python, and Go examples), and troubleshooting. Opt-in per organization.
Navigate to changelog: 25th May 2026
Copilot now returns
searchQueries
(query fan-out)
/v1/monitor/copilot
responses now include
result.searchQueries
— the web search queries Copilot ran while generating its answer. Always-on, no
include
flag, no surcharge.
"searchQueries"
: [
"best laptops for programming 2026"
,
"developer laptop reviews"
]
Navigate to changelog: 22nd May 2026
Structured
citationPills
now on Perplexity, Copilot, and Gemini
All three endpoints now emit an additive
citationPills
array alongside the answer text and sources. Combined with the May 21st rollout to Google AI Overview and AI Mode, cross-provider iteration over AI citation pills now works the same way on six endpoints (ChatGPT, Google AI Overview, AI Mode, Perplexity, Copilot, Gemini).
Each entry is one (pill, source) pair carrying a per-source
label
(the source’s own page title), a
citationPillId
that groups entries from the same chip, and per-source
url
/
domain
/
description
/
position
.
"citationPills"
: [
{
"label"
:
"Best Programming Laptops 2026 — TechCrunch"
,
"citationPillId"
:
1
,
"url"
:
"https://techcrunch.com/laptops"
,
"domain"
:
"techcrunch.com"
,
"description"
:
"Latest laptop reviews and recommendations"
,
"position"
:
1
},
{
"label"
:
"Wired Laptop Buyer's Guide"
,
"citationPillId"
:
1
,
"url"
:
"https://wired.com/programming-laptops"
,
"domain"
:
"wired.com"
,
"description"
:
"Programming laptop buyer's guide"
,
"position"
:
3
}
]
All six providers share a common base shape (
label
,
citationPillId
,
url
,
domain
,
description?
,
position
). ChatGPT carries two additional fields (
datePublished?
,
type
) that the other surfaces do not emit. Consumers writing cross-provider citation-pill iteration should treat
datePublished
and
type
as ChatGPT-only, and use
citationPillId
for chip grouping.
Additive change — existing consumers are unaffected. The field is omitted from
result
when the answer carries no pills. Learn more in the per-endpoint citation pills documentation:
Perplexity
,
Copilot
,
Gemini
.
Navigate to changelog: 21st May 2026
Structured
citationPills
array on Google AI Overview and AI Mode
Both Google AI surfaces now emit an additive
citationPills
array alongside the answer text and sources. Each entry is one (pill, source) pair carrying a per-source
label
(the source’s own page title from the sources rail), a
citationPillId
that groups entries from the same chip, and per-source
url
/
domain
/
description
/
position
. Cross-provider iteration over AI surface citation pills now uses one shape.
"citationPills"
: [
{
"label"
:
"Best laptops for programmers in 2026"
,
"citationPillId"
:
1
,
"url"
:
"https://example.com/best-laptops-2026"
,
"domain"
:
"example.com"
,
"description"
:
"A guide to the top laptops for software development this year."
,
"position"
:
1
},
{
"label"
:
"MacBook Pro for developers"
,
"citationPillId"
:
1
,
"url"
:
"https://example.com/macbook-pro-review"
,
"domain"
:
"example.com"
,
"description"
:
"Why the MacBook Pro is a popular pick for developers."
,
"position"
:
3
}
]
When a pill cites N sources, the array contains N entries sharing the same
citationPillId
but carrying distinct per-source
label
,
url
, and
domain
. Group by
citationPillId
to recover the pill-level structure.
Markdown rendering is unchanged — inline
[anchor](url)
links remain the canonical in-text citation form. The
citationPills
array gives type-aware consumers per-source access without re-parsing markdown.
Additive change — existing consumers are unaffected. Learn more in the
AI Overview citation pills
and
AI Mode citation pills
documentation.
Navigate to changelog: 19th May 2026
AI Overview ads now carry a
type
discriminator
Every entry in
aioverview.ads[]
on the Google Search endpoint now carries an explicit
type
field. Mirrors the same convention already in place for top-level
ads[]
(added 13th May), so callers can route ads to per-sub-type rendering without re-implementing the price/store presence heuristic.
type: "TEXT" — text/lead-gen ads. Carry domain + description (+ shared position / title / url / image ).
type: "SHOPPING" — shopping/product ads. Carry price + store (+ optional old_price , + shared fields).
Each ad is exactly one sub-type — the two sub-schemas are disjoint. Sub-type-exclusive fields are only present when they apply to that ad, matching what the Google SERP actually renders.
image
is the one field that appears in either sub-type (product photo on shopping cards, hero image on text ads), so consumers should treat it as optional regardless of
type
.
"ads"
: [
{
"position"
:
1
,
"type"
:
"TEXT"
,
"title"
:
"..."
,
"url"
:
"..."
,
"domain"
:
"dell.com"
,
"description"
:
"..."
},
{
"position"
:
2
,
"type"
:
"SHOPPING"
,
"title"
:
"..."
,
"url"
:
"..."
,
"price"
: {
"value"
:
1499
,
"currency"
:
"$"
},
"store"
:
"Best Buy"
}
]
Additive change — existing consumers ignoring the new field are unaffected. No opt-in, no extra credit charge. Learn more in the
AI Overview ads documentation
.
Navigate to changelog: 19th May 2026
AI Overview and AI Mode markdown citations now match ChatGPT
When you opt into markdown output, in-text citations from Google AI Overview and Google AI Mode are now rendered as
standard inline
[anchor](url)
links
in the
markdown
field — the same shape ChatGPT already uses.
... the iPhone 17 is the best choice for most people.[CNET](https://www.cnet.com/tech/mobile/best-iphone/)\[The best iPhone to buy!!](https://www.youtube.com/shorts/z1aZ3CYcaRE)
Anchor text resolves via a deterministic fallback chain: matched source title → Google’s pill-chip label → URL host. Multi-source pills (Google’s
+N
chips) expand to N adjacent stacked links at the citation point.
result.markdown
is now fully self-contained — no parallel index resolution needed to render citations.
Learn more in the
AI Overview
and
AI Mode
documentation.
Navigate to changelog: 18th May 2026
AI Overview videos now carry full metadata
aioverview.videos[]
entries on the Google Search endpoint now populate
title
,
source
,
platform
,
date
,
thumbnail
, and
duration
whenever Google renders them. Previous responses returned only
url
for the vast majority of video cards because the extractor was matching stale Google class names.
url
remains the only guaranteed field.
title
,
source
,
platform
, and
date
are present on roughly 98% of videos;
thumbnail
(~60%) and
duration
(~15%) appear only when Google renders the rich carousel preview. Check for field presence before reading.
No schema change, no opt-in, no extra credit charge — existing requests start receiving the richer payload automatically. Learn more in the
AI Overview documentation
.
Navigate to changelog: 13th May 2026
Google Search now returns sponsored shopping-card ads
The Google Search endpoint now extracts every sponsored shopping surface Google renders on the SERP, alongside the existing text ads. Each entry in
ads[]
carries a new
type
discriminator:
type: "RESULT" — classic text ads at the top or bottom of the page (unchanged).
type: "SHOPPING_CARD" — shopping-style sponsored cards from the right-hand-side carousel ( blockPosition: "rhs" ) and the top-of-page carousels ( blockPosition: "top" ). Categories observed include "Sponsored products" , "Sponsored vehicles" , and "Sponsored hotels" .
SHOPPING_CARD
ads include
category
(the carousel header label),
price
, optional
oldPrice
(MSRP / sale),
store
, and
imageUrl
. For shopping-card ads, the existing
description
field carries category-specific subtitle fragments joined with
·
(e.g.
Used - 94k miles · Greeley
on a vehicle card).
The
ads[]
blockPosition
enum widens to
"top" | "bottom" | "middle" | "rhs"
. The new fields are emitted whenever the SERP renders the corresponding surface — no opt-in, no additional credit charge.
In the organic
shoppingCards[]
field, the
product_link
and
old_price
keys are now emitted as camelCase
productLink
and
oldPrice
to match the rest of the Google response.
Learn more in the
sponsored ads documentation
.
Navigate to changelog: 11th May 2026
Google Search and Google News multi-page pricing is now linear
Multi-page requests are now billed at
+2 credits per additional page
instead of the previous tiered structure. Single-page requests are unchanged.
| Request | Old cost | New cost |
|---|---|---|
| 1 page | 3 credits | 3 credits (unchanged) |
| 1 page + AI Overview | 5 credits | 5 credits (unchanged) |
| 3 pages | 4 credits | 7 credits |
| 3 pages + AI Overview | 6 credits | 9 credits |
| 10 pages | 5 credits | 21 credits |
| 10 pages + AI Overview | 7 credits | 23 credits |
The same per-page rate applies to
Google News
. See
providers
for the full pricing breakdown.
Why the change.
Pages 2-10 are materially harder to serve than page 1: the failure rate climbs as Google ramps up scrutiny on deep pagination. In our infrastructure, fetching pages 1-10 of a single query costs roughly
6× more
than fetching 10× page 1 alone — the per-page cost scales near-exponentially with depth. The old tiered pricing absorbed that cost and under-charged deep-pagination requests; the new per-page rate aligns billing with the actual cost of serving each page and keeps single-page Google requests — the dominant use case — unchanged.
Navigate to changelog: 10th May 2026
Grok pricing increases to 4 credits per request
Grok’s anti-bot defenses have grown materially more complex over recent weeks, raising the proxy, fingerprinting, and retry cost of every successful extraction. To keep base pricing aligned with the underlying infrastructure cost, the
Grok endpoint
now costs
4 credits per request
(previously 3). All other provider prices, addons, and the sync surcharge are unchanged. See the full
providers pricing table
for context.
Navigate to changelog: 7th May 2026
Google Search now returns shopping cards and “people are saying” cards
The Google Search endpoint now extracts two additional SERP modules:
Shopping cards : product cards from the organic shopping grids (“Popular products” and “More products”) in a new optional shoppingCards array. Learn more in the shopping cards documentation .
People are saying : community-thread cards from Google’s “What people are saying” / “Trending posts and discussions” module in a new optional peopleAreSaying array. Learn more in the people are saying documentation .
Both fields are omitted from
result
when their corresponding module is absent — treat them as optional. No additional credit charge.
Navigate to changelog: 5th May 2026
Gemini sources no longer include
confidence_level
Google Gemini stopped returning a confidence score on source citations. The
confidence_level
field has been removed from the
Gemini endpoint
response schema. All other source fields (
position
,
url
,
label
,
description
) are unchanged.
Navigate to changelog: 29th April 2026
More Google AI Mode fields
The AI Mode endpoint now returns more structured data:
Map : GPS-enriched location results with gpsCoordinates (latitude/longitude), type , thumbnail , and a position index
Places : Inline place cards with rating , reviews , type , priceLevel , address , and status
Shopping cards : Shopping cards now include oldPrice for discount comparison, snippet for product descriptions, and snippetLinks for related links within snippets
Ads : Sponsored ad sections are now parsed into a structured ads object with individual ad details including title, URL, position, price, store, rating, and reviews
Inline products : Product cards embedded within the AI response text are now extracted as inlineProducts , separate from shopping card carousels
Breaking change
: The
price
and
description
fields have been removed from the places object. The
link
field now points to Google’s viewer URL format instead of the previous search URL format.
All new data is included at no additional credit charge. Learn more in the
AI Mode endpoint documentation
.
Navigate to changelog: 25th April 2026
Inline products now return offers correctly for ChatGPT
We fixed a bug where ChatGPT inline products were missing pricing and merchant data. Inline products now include the full offer payload (current price, multi-merchant offers from retailers like Amazon and John Lewis, product images, and aggregated ratings) fetched from ChatGPT’s product update API.
Alongside this fix, shopping cards and inline products move to an opt-in model on the ChatGPT endpoint. Set
include.shopping: true
to include them in your response:
Pricing and offers : Current price and multi-merchant offers with availability
Product images : Image URLs for product display
Ratings : Aggregated rating scores and review counts
Opt-in with additional credit cost
Shopping cards and inline products are disabled by default. Set
include.shopping: true
to enable them. Enabling
shopping
,
rawResponse
,
searchQueries
, or
ads
(or any combination) adds +2 credits to the base cost.
The credit surcharge reflects increased extraction complexity. OpenAI now serves product pricing, offers, and merchant data through a separate endpoint, so cloro performs additional fetches per product to enrich the response.
Migration deadline: 8th May 2026
During the migration window, ChatGPT responses still return
shoppingCards
and
inlineProducts
by default.
Starting 8th May 2026
, requests without
include.shopping: true
will no longer receive these fields. Update your integrations before that date to avoid silent data loss.
Breaking change
: Shopping cards and inline products were previously included automatically in ChatGPT responses. They now require
include.shopping: true
in your request. The
details
object on inline products (rationale and themed reviews) is no longer returned.
Learn more in our
ChatGPT shopping cards documentation
and
inline products documentation
.
Navigate to changelog: 23rd April 2026
Hyperlocal-targeting for AI Mode with UULE
The AI Mode endpoint now supports two parameters for precise geo-targeting of results:
location : City-level targeting using Google canonical location names (e.g., New York,New York,United States ). Choose from ~100,000 supported locations worldwide
uule : Advanced targeting using pre-encoded Google UULE strings for users who generate their own UULE values
Both parameters work alongside
country
for precise localization. They are mutually exclusive: provide one or the other, not both. No additional credits required.
Learn more in our
AI Mode documentation
.
Navigate to changelog: 23rd April 2026
Copilot pricing update
The Copilot endpoint base cost has been updated to 5 credits per request, to reflect the added complexity to ensure sources in 95%+ of cases.
Navigate to changelog: 23rd April 2026
Sync request surcharge
All sync monitor requests (
/v1/monitor/*
) now include a
+2 credit surcharge
on top of the base cost and any feature add-ons. This reflects the higher operational cost of real-time synchronous requests.
Async and batch requests (
/v1/async/*
) are
not
affected by this surcharge.
Existing customers who used sync endpoints in the 30 days before this change have been granted a grace period. If you have questions about your account, contact
.
Learn more in our
provider pricing documentation
.
Navigate to changelog: 22nd April 2026
Geo-targeting for Google Search
The Google Search endpoint now supports two parameters for precise geo-targeting of search results:
location : City-level targeting using Google canonical location names (e.g., New York,New York,United States ). Choose from ~100,000 supported locations worldwide
uule : Advanced targeting using pre-encoded Google UULE strings for users who generate their own UULE values
Both parameters work alongside
country
for precise localization. They are mutually exclusive — provide one or the other, not both. No additional credits required.
Learn more in our
Google Search documentation
.
Navigate to changelog: 17th April 2026
Sponsored ad extraction for AI Overview
Google AI Overview responses now include structured sponsored ad data when Google injects advertising inside the AI Overview. Ads are exposed as a new
ads
array alongside the existing
text
,
sources
, and
videos
fields.
Each ad includes:
Ad details : Title, destination URL, domain, and description
Position tracking : Position index within the AI Overview ad block
Ads also remain visible in
text
and
markdown
output for full transparency. No additional parameters required — ads are automatically included when present.
Learn more in our
AI Overview documentation
.
Navigate to changelog: 15th April 2026
Map entries for Copilot
The Copilot endpoint now automatically extracts business and place data when Copilot returns local entity information. Map entries include location coordinates, reviews, photos, open/closed status, and Google Place IDs. No additional parameters required.
Features
:
Automatic extraction : Map entries are included by default when available, just like shopping cards
Native structure : Preserves Copilot’s entity format with nested location , reviews , and photos objects
Google Places integration : Each entry includes a placeId for direct Google Maps lookups
Live status : openState field shows current business hours (e.g. “Open · Closes 9 PM”)
No extra cost : Map entries are included at no additional credit charge
Learn more in the
Copilot endpoint documentation
.
Navigate to changelog: 9th April 2026
Batch task creation
You can now submit up to 500 async tasks in a single API request using the new batch endpoint. This reduces HTTP overhead for high-volume workloads and simplifies bulk monitoring workflows.
Features
:
Up to 500 tasks per request : Submit tasks to any combination of providers in one call
Partial success : Each task is validated independently. One invalid task doesn’t block the rest
Per-task error reporting : Failed tasks include detailed error codes ( VALIDATION_ERROR , RESOURCE_ALREADY_EXISTS , INSUFFICIENT_CREDITS ) with field-level details
Per-task credits : Each task’s result includes its own credit information
Full feature support : Each task in a batch supports priority, idempotency keys, and webhooks
Learn more in the
batch task creation documentation
.
Navigate to changelog: 9th April 2026
Priority support for async tasks
You can now assign a priority to async tasks so that time-sensitive work is processed first. Set the
priority
field (1-10) when creating a task. Higher numbers mean higher urgency.
Features
:
Priority range : Integer from 1 (lowest, default) to 10 (highest)
Smart ordering : Higher-priority tasks are processed before lower-priority ones within your queue. Same-priority tasks are processed in FIFO order
Priority breakdown : The async status endpoint now shows queued task counts per priority level
Backward compatible : Tasks without a priority default to 1, so existing integrations are unaffected
Learn more in our
async requests documentation
.
Navigate to changelog: 7th April 2026
More business data in ChatGPT map entries
ChatGPT map entries now return mapped business data from providers (Yelp, Google Business) with 47 structured fields, replacing the previous 10-field structure.
What changed:
Mapped data : Map entries now include all available business fields from ChatGPT’s providers (Yelp, Google Business, Yelp-feed)
47 mapped fields : Access to detailed location data, operating hours, attributes, reviews, images, and provider-specific metadata
Structured schema : Properly typed fields with consistent camelCase naming conventions
Field name updates : phone_number → phone , country → countryCode , review_count → reviewCount
Null-safe : All fields except name and position are optional
Key improvements:
More location data : Full address breakdown with coordinates ( latitude / longitude ), distance from user ( distanceMeters ), and human-readable location strings
Hours : Weekly schedule with next opening times ( nextOpenHour ) and special hours ( specialHours )
Attributes : Provider-specific details like parking options, WiFi availability, accepted payment methods, and accessibility features
Images : Multiple image URLs ( imageUrls ) with provider-hosted images ( providerImages )
Metadata : Business claimed status ( isClaimed ), opening/closing dates ( dateOpened / dateClosed ), popularity scores ( popularityScore ), and cache indicators ( fromCache )
Breaking change
: Field names have changed to camelCase. Examples:
review_count
→
reviewCount
,
country_code
→
countryCode
,
is_open
→
isOpen
. All fields except
name
and
position
are now optional.
Learn more in our
ChatGPT map documentation
.
Navigate to changelog: 6th April 2026
Ad extraction for ChatGPT
ChatGPT responses now support structured ad extraction when advertising content is displayed. This opt-in feature lets you monitor ads shown in ChatGPT responses.
Opt-in feature with additional cost
Ad extraction is disabled by default. Set
include.ads: true
to extract ads from ChatGPT responses. Enabling
ads
,
rawResponse
, or
searchQueries
(or any combination) adds +2 credits to the base cost.
Features
:
Brand information : Advertiser brand name, URL, and favicon
Carousel cards : Multiple promotional cards with titles, descriptions, images, and destination URLs
Attribution tracking : All URLs include ChatGPT attribution parameters
Structured data : JSON structure
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 1st April 2026
Google News endpoint
We’ve added a new endpoint for extracting structured news articles from Google News (
/v1/monitor/google/news
).
Features
:
News article extraction : Titles, links, snippets, sources, and publication dates
Multi-page scraping : Scrape up to 10 pages of news results per request
Country-specific results : Localized news from 250 countries worldwide
Device targeting : Desktop or mobile news results
Thumbnail images : Article thumbnails when available
Raw HTML access : Optional full page HTML for custom parsing
Learn more in our
Google News documentation
.
Navigate to changelog: 30th March 2026
Inline products for ChatGPT
ChatGPT responses now include inline products: individual product references that appear embedded in the response text, separate from shopping cards. This feature supports OpenAI’s new
product discovery capabilities
announced in ChatGPT.
Model availability
: Inline products currently appear when ChatGPT uses the
gpt-5-3
model. ChatGPT automatically selects which model to use for each request.
What’s new:
Inline product extraction: Individual products referenced in comparison tables, featured recommendations, or inline mentions
Rendering hints: Display guidance with render_as field (“inline”, “hero”, or “block”)
Cite references: Each product includes a unique cite ID for cross-referencing with text
Learn more in our
ChatGPT inline products documentation
.
Navigate to changelog: 24th March 2026
Async queue monitoring and visibility
New
GET /v1/async/status
endpoint provides real-time visibility into your async queue health, including queued tasks, processing tasks, and concurrency usage.
Use cases
:
Monitor queue health and processing capacity
Decide on plan upgrades based on actual concurrency usage
Debug task delays by identifying queue bottlenecks
Throttle task submission based on current queue size
Learn more in our
async status documentation
.
Navigate to changelog: 20th March 2026
Shopping cards for Copilot
Copilot responses now automatically include structured shopping cards when product information is detected:
Product identifiers : Unique product ID, group ID, and brand group ID for precise tracking
Product details : Name, description, brand, and seller information
Specifications : Configurable product attributes like Color, Size with available values
Images : Multiple product images with titles
Pricing : Structured price data with amount, currency, and symbol
Ratings : Product ratings with review counts
Price tracking : Flag indicating if price tracking is available
No additional parameters are required. Shopping cards are automatically detected and included when Copilot returns product information.
Learn more in our
Copilot documentation
.
Navigate to changelog: 19th March 2026
Raw response now available for all event stream engines
The
rawResponse
field is now available for Copilot, Grok, Gemini, and Perplexity endpoints at no additional cost.
Example usage:
{
"prompt"
:
"Your query here"
,
"model"
:
"GROK"
,
"include"
: {
"rawResponse"
:
true
}
}
Raw response data provides full streaming event payloads, so you can see how these AI models generate their responses.
Navigate to changelog: 14th March 2026
Global coverage milestone: 250 countries supported
cloro now supports
250 countries worldwide
across all endpoints.
Our geographical infrastructure spans from major markets to remote territories, so you can monitor AI responses and search results from almost anywhere in the world. This includes territories like Antarctica (AQ), remote islands like Bouvet Island (BV), and regions like the British Indian Ocean Territory (IO).
Total coverage
:
250 countries : Complete ISO 3166-1 alpha-2 country code coverage
All endpoints supported : Geographical availability across ChatGPT, Perplexity, Grok, Gemini, Copilot, AI Mode, AI Overview, and Google Search
Model-specific availability : Each model has its own country support based on provider restrictions
The
countries endpoint
remains the authoritative source for current country availability. Always query with your target model to verify supported locations.
Navigate to changelog: 6th March 2026
Massive geographical expansion for Perplexity, Copilot, Gemini, and Grok
We’ve expanded country coverage for Perplexity, Copilot, Gemini, and Grok to match the availability of ChatGPT.
Country coverage by model
:
ChatGPT : 196 countries (unchanged)
Perplexity : 196 countries (expanded from 74)
Grok : 195 countries (expanded from 74)
Gemini : 195 countries (expanded from 73)
Copilot : 193 countries (expanded from 71)
AI Mode : 211 countries (unchanged)
AI Overview : 230 countries (unchanged)
Google Search : 250 countries (unchanged)
The
countries endpoint
remains the authoritative source for current country availability. Always query with your target model to verify supported locations before deploying.
Navigate to changelog: 3rd March 2026
Grok search queries and model extraction
Grok responses now include two new fields for better insight into how responses are generated:
searchQueries : Array of search queries Grok used to gather information for the response
model : The model identifier used to generate the response (e.g., “grok-3”, “grok-4-auto”)
Example response
:
{
"result"
: {
"text"
:
"Here are the best sneakers under $100..."
,
"searchQueries"
: [
"best sneakers under $100 2026"
,
"best budget sneakers under $100 2026"
],
"model"
:
"grok-4-auto"
}
}
Learn more in our
Grok documentation
.
Navigate to changelog: 26th February 2026
Copilot country restrictions
Italy (IT) and Brazil (BR) have been disabled for the Copilot endpoint. Microsoft has placed these regions behind a login wall, making them unavailable for anonymous scraping.
Use the
countries endpoint
with
model=copilot
to verify current country availability before deploying.
Navigate to changelog: 19th February 2026
Improved ChatGPT source metadata
ChatGPT sources now include enhanced metadata for better source attribution:
datePublished : Publication date of the source article (e.g., “May 22, 2025”)
label : Now contains the article title instead of the domain name
description : Now contains a content snippet instead of the combined date and title
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 10th February 2026
ChatGPT geographical expansion
ChatGPT is now available in 200 countries worldwide.
The
countries endpoint
remains the authoritative source for current country availability. Always query with
model=chatgpt
to verify supported locations before deploying.
Navigate to changelog: 9th February 2026
Sponsored ad extraction for Google Search
Google Search endpoint now automatically extracts sponsored ad results from search results pages.
Features
:
Ad placement tracking : Ads are extracted from both top and bottom of search results with blockPosition indicating placement
Complete ad details : Title, URL, displayed URL, domain, description, and position within ad block
Ad sitelinks : Extended ad sitelinks with titles, URLs, and descriptions when available
Multi-page support : Ads are extracted across all requested pages
Response structure
:
{
"result"
: {
"ads"
: [
{
"position"
:
1
,
"blockPosition"
:
"top"
,
"title"
:
"Best Programming Laptops - Shop Now"
,
"url"
:
"https://example.com/programming-laptops"
,
"page"
:
1
,
"displayedUrl"
:
"example.com/laptops"
,
"domain"
:
"example.com"
,
"description"
:
"Shop our selection of high-performance laptops..."
,
"sitelinks"
: [
{
"url"
:
"https://example.com/gaming-laptops"
,
"title"
:
"Gaming Laptops"
,
"description"
:
"High-performance laptops for gaming"
}
]
}
]
}
}
No additional parameters required. Ads are automatically included when present on search results pages.
Learn more in our
Google Search documentation
.
Navigate to changelog: 6th February 2026
Isolation of ChatGPT citation data
We’ve improved how source information is structured and exposed in ChatGPT responses:
Citation pills
: Inline citations that appear within ChatGPT responses where specific sources are referenced. Each citation pill includes:
URL : Direct link to the cited source
Label : Title or label of the citation
Description : Summary of the cited content
Domain : Source domain (e.g., “example.com”)
Date published : ISO 8601 date string when the source was published
Citation pill ID : Unique identifier for the citation
Sources footnote field
: The
sources
array now includes a
footnote
boolean field that indicates whether a source appears in the main window footnote (i.e., the sources pill), to distinguish between primary and secondary sources.
Previously, the citation and source pill data was only accessible by parsing the markdown response. Now you can access it directly through structured JSON objects without needing to parse markdown links.
No additional parameters required. Citation pills and the footnote field are automatically included when available in ChatGPT responses.
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 5th February 2026
Updated Google Search pricing structure
We’ve updated the Google Search endpoint baseline pricing to a more predictable tiered structure for multi-page requests.
What’s changed:
Multi-page pricing is now tiered instead of per page: 2-3 pages: +1 credit (was +2-4 credits) 4-10 pages: +2 credits (was +6-18 credits)
Base cost and AI Overview remain unchanged: Base request: 3 credits AI Overview addon: +2 credits
New pricing examples:
1 page, no AI Overview: 3 credits (unchanged)
1 page, with AI Overview: 5 credits (unchanged)
3 pages, with AI Overview: 6 credits (was 9 credits)
10 pages, with AI Overview: 7 credits (was 23 credits)
Costs are more predictable, with reduced pricing for bulk multi-page extraction.
Navigate to changelog: 3rd February 2026
Map entries for ChatGPT
ChatGPT responses now automatically include structured map entries when business or place information is detected. Use cases:
Local business monitoring : Track restaurants, stores, and service providers
Location intelligence : Extract ratings, reviews, and contact details
Competitive analysis : Monitor local businesses across different regions
Directory building : Create structured business listings
Map entries include business details:
Business information : Name, category, description
Social proof : Rating score, review count
Contact details : Website URL, phone number
Location data : Position/ranking in results
Navigation : Directions URL (when available)
No additional parameters are required. Map entries are automatically detected and included when ChatGPT returns local business or place information.
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 31st January 2026
Added 16 new countries
We’ve expanded our geographical coverage with 16 new countries available for monitoring across all non-Google endpoints:
BG - Bulgaria
BY - Belarus
CY - Cyprus
EE - Estonia
GE - Georgia
KE - Kenya
KG - Kyrgyzstan
LT - Lithuania
LU - Luxembourg
LV - Latvia
MA - Morocco
MT - Malta
PE - Peru
TJ - Tajikistan
UZ - Uzbekistan
Remember to always check the
model-specific country list
before deploying to ensure availability for your target endpoint, as some models have country-specific restrictions.
Navigate to changelog: 23rd January 2026
Google Search and AI Mode updates
We’ve updated the Google Search and AI Mode endpoints:
Removed parameters:
city parameter temporarily removed from Google Search and AI Mode endpoints
No customers were using this parameter, and country -level localization was sufficient for all use cases
New parameter:
device parameter added to AI Mode endpoint
Options: desktop (default) or mobile
Same device targeting available in Google Search
Navigate to changelog: 20th January 2026
Gemini geographical expansion
Gemini now supports EU countries and additional regions, bringing geographical coverage to parity with Copilot, Perplexity, and Grok.
The
countries endpoint
remains the authoritative source for current country availability. Always query with
model=gemini
to verify supported locations before deploying:
curl
"https://api.cloro.dev/v1/countries?model=gemini"
\
-H
"Authorization: Bearer YOUR_API_KEY"
Navigate to changelog: 15th January 2026
Grok is now available
Grok is now available as a new provider.
Key features:
Supported in most countries (remember to check the model-specific country list )
Source metadata with preview text, search engine display text, site information, creator details, and image URLs
Real-time web integration for current information and breaking news
Response formats : text, HTML, and Markdown
Getting started:
curl
-X
POST
https://api.cloro.dev/v1/monitor/grok
\
-H
"Authorization: Bearer YOUR_API_KEY"
\
-H
"Content-Type: application/json"
\
-d
'{
"prompt": "What are the latest developments in quantum computing?",
"country": "US"
}'
See the
Grok API documentation
for full details and examples.
Navigate to changelog: 11th January 2026
We’ve expanded our geographical coverage across all Google endpoints:
Google Search : Now available in all 250 countries worldwide
AI Mode : Expanded from 58 to 211 countries
AI Overview : Expanded from 59 to 230 countries
Remember to check the
model-specific country list
before deploying to ensure availability for your target endpoint.
Navigate to changelog: 18th December 2025
Added 11 new countries
Expanded our geographical coverage with 11 new countries available for monitoring:
BD - Bangladesh
CO - Colombia
EG - Egypt
FI - Finland
HK - Hong Kong
KZ - Kazakhstan
NG - Nigeria
PH - Philippines
PK - Pakistan
VE - Venezuela
VN - Vietnam
Most new locations start with low capacity, which will be increased based on demand. Remember to check the
model-specific country list
before deploying to ensure availability for your target endpoint.
Navigate to changelog: 17th December 2025
Defined async task retention policy
COMPLETED
and
FAILED
async tasks are stored for
24 hours
after you create the task.
HTML URLs also expire after 24 hours from generation (aligned with task retention).
Navigate to changelog: 16th December 2025
More fields extracted for AI Mode and AI Overview
Google AI Mode and AI Overview endpoints now automatically extract rich structured data:
Places in AI Mode : Location information with ratings, reviews, price levels, addresses, Google Places IDs, and operating status
Shopping cards in AI Mode : Structured product data with currency-aware pricing, merchant details, ratings, and purchase links
Video content in AI Overview : Videos with thumbnails, titles, duration, platform info, and source attribution
No additional parameters required. All structured data is automatically detected and included when available in responses.
Learn more in
AI Mode
and
AI Overview
.
Navigate to changelog: 12th December 2025
Since location availability depends on the model, we have enhanced the
/v1/countries
endpoint with model filtering:
Model-specific filtering - Get countries available for specific AI providers using the model query parameter
Supported models - Filter by: aimode , aioverview , chatgpt , copilot , gemini , google , or perplexity
Usage examples
:
# Get all countries
curl
"https://api.cloro.dev/v1/countries"
# Get countries available for ChatGPT
curl
"https://api.cloro.dev/v1/countries?model=chatgpt"
# Get countries available for Perplexity
curl
"https://api.cloro.dev/v1/countries?model=perplexity"
Learn more in our
countries endpoint documentation
.
Navigate to changelog: 23rd November 2025
We’ve added support for Google Gemini (
/v1/monitor/gemini
). You can now extract structured data from Gemini responses with source citations.
Features
:
Text extraction - Get the generated response text
Source citations - Structured list of sources with URLs, snippets, and confidence levels
Markdown & HTML - Optional output formats
Usage example
:
{
"prompt"
:
"Explain quantum computing"
,
"country"
:
"US"
,
"include"
: {
"markdown"
:
true
}
}
Learn more in our
Gemini documentation
.
Navigate to changelog: 18th November 2025
The HTML code is now stored externally to avoid crowding the response payload.
Changes
:
HTML URLs now expire after 24 hours from generation
Storage URLs updated to use https://storage.cloro.dev/results/ format
Important note
: If you need to preserve HTML content longer than 24 hours, download it promptly after receiving the response.
Example response
:
{
"result"
: {
"html"
:
"https://storage.cloro.dev/results/c45a5081-808d-4ed3-9c86-e4baf16c8ab8/page-1.html"
}
}
Learn more in our
output formats documentation
.
Navigate to changelog: 17th November 2025
New Google Search endpoint
We’re excited to announce the launch of our Google Search endpoint (
/v1/monitor/google
) for extracting structured data from Google search results.
Features
:
Organic results - Extract titles, links, snippets, and positions from search results
People Also Ask - Capture questions and answers with page tracking
Related searches - Get related search query suggestions
AI Overview - Extract Google’s AI-generated summaries with source attribution (optional)
Multi-page scraping - Scrape up to 10 pages per request with the pages parameter
Country-specific results - Localized search results across many countries
Hyperlocal targeting - City-specific search results using canonical city names with automatic UULE conversion
Raw HTML access - Option to retrieve full page HTML for custom parsing
Usage example
:
{
"query"
:
"best laptops for programming"
,
"pages"
:
3
,
"country"
:
"US"
,
"include"
: {
"aioverview"
: {
"markdown"
:
true
}
}
}
Learn more in our
Google search documentation
.
Navigate to changelog: 15th November 2025
Enhanced retry logic for synchronous requests
We have the following changes in our internal retry mechanism:
Increased retry attempts from 5 to 10 for synchronous requests
Added 5-minute maximum duration timeout to prevent indefinite waiting
Requests canceled by clients are charged for resources consumed (same behavior as before)
This automatic retry handling
removes the need for client-side timeout logic
. More details in the
Error handling
section.
Navigate to changelog: 14th November 2025
Configurable HTML response for all parsers
HTML content in parser responses is now optional and configurable across all monitoring endpoints.
Changes
:
Added include.html parameter to all parser endpoints (ChatGPT, Perplexity, Copilot, AI Mode, AI Overview)
A URL to the HTML content is now only included when explicitly requested with "include.html": true
Responses are smaller and faster when HTML is not needed
Markdown conversion continues to work regardless of HTML inclusion
Usage example
:
{
"prompt"
:
"Your query here"
,
"include"
: {
"html"
:
true
,
"markdown"
:
true
}
}
This reduces response size and improves performance when your application doesn’t need the full HTML content.
Navigate to changelog: 11th November 2025
Asynchronous requests and webhooks
You can now make asynchronous requests for async tasks. When you make an async request, you’ll receive a
taskId
, the task will then be processed in the background.
You can retrieve the results in two ways:
Webhooks : Provide a webhook.url in your request, and we’ll send the results to your endpoint as soon as they’re ready.
Polling : Use the new GET /v1/async/task/{taskId} endpoint to check the status of your task and retrieve the results when it’s complete.
This works for serverless environments or applications that can’t wait for a task to complete.
Learn more in our
async requests documentation
.
Navigate to changelog: 7th November 2025
More information in Perplexity response
The Perplexity API endpoint has been expanded to reflect extra fields in the Perplexity response. The endpoint now automatically extracts structured data objects based on query intent:
Shopping data extraction
:
Product information with titles, descriptions, pricing, and ratings
Multiple merchant offers and availability status
Product variants, images, and specifications
Shopping cards with promotional tags and categorization
Media content extraction
:
Video content with thumbnails, dimensions, and source attribution
Image extraction with metadata and sizing information
Support for YouTube, stock photos, and other media platforms
Travel and location data
:
Hotel listings with ratings, reviews, amenities, and pricing
Place information with categories, coordinates, and contact details
Address data, map URLs, and image galleries
Search intelligence
:
Related query suggestions for follow-up searches
Internal search query tracking showing how responses are generated
Citation processing and source attribution
The documentation now provides schema definitions for all response objects, including field types, descriptions, and examples. No API changes were required. These features were already available and are now documented.
Learn more in the updated
Perplexity API documentation
.
Navigate to changelog: 6th November 2025
Entity extraction for ChatGPT
ChatGPT responses now include automatic entity extraction when specific items, products, brands, or concepts are identified. Use cases:
Product recognition : Identify and extract product mentions
Brand tracking : Monitor brand references across ChatGPT responses
Concept analysis : Extract named entities for semantic analysis
Content classification : Categorize response content based on extracted entities
Each entity includes:
Type : Entity type identifier (e.g., “product”, “software”)
Name : Entity name or title (e.g., “adidas Grand Court Lo”, “Nike Dunk Low Retro SE”)
Entity extraction works alongside shopping cards. No additional parameters are required. Entities are automatically detected and included when available.
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 1st November 2025
Shopping cards for ChatGPT
ChatGPT responses now automatically include structured shopping cards when product information is detected. Use cases:
Product monitoring : Track prices, ratings, and availability over time
Price tracking : Monitor price changes across different regions
Competitive intelligence : Extract structured product data for analysis
Commerce integration : Build price comparison and recommendation systems
Shopping cards include product details:
Product information : Name, brand, description, specifications
Pricing data : Current price, original price, discount information
Commercial details : Multiple merchant offers, availability, checkout options
Media assets : Product images, checkout-specific images
Ratings : Ratings, review counts, rating citations with sources
Offer details : Merchant-specific pricing, promotional tags, shipping costs
The shopping cards feature supports:
Multi-offer products : Compare different merchants for the same product
Price breakdown : Pricing including base, shipping, tax, and total
Promotional content : Tags, tooltips, and special offers
Media : Multiple product images and checkout assets
Availability tracking : Real-time stock and checkout status
No additional parameters are required. Shopping cards are automatically detected and included when available.
Learn more in our
ChatGPT documentation
.
Navigate to changelog: 31st October 2025
New dedicated endpoints
Each AI model now has its own dedicated endpoint for better performance and clearer documentation:
Extract Google AI Overview (via POST /v1/monitor/google with include.aioverview )
Extract Microsoft Copilot ( POST /v1/monitor/copilot )
Extract Perplexity ( POST /v1/monitor/perplexity )
Extract Google AI Mode ( POST /v1/monitor/aimode )
Extract ChatGPT ( POST /v1/monitor/chatgpt )
Navigate to changelog: 18th October 2025
Added Google AI Mode
Google’s AI Mode results are now supported. You can extract structured data from ChatGPT, Perplexity, Copilot, and AI Mode using the same unified interface.
To monitor AI Mode, set
"model": "AIMODE"
in your API requests:
{
"prompt"
:
"What do you know about Acme Corp?"
,
"model"
:
"AIMODE"
,
"country"
:
"US"
}
All existing features work with AIMODE, including:
Structured data extraction with sources & citations
Localization across different countries
Markdown format support
Citation grouping and processing
Learn more in our
API Reference
.
Navigate to changelog: 17th October 2025
Added
include.rawResponse
parameter
You can now access the underlying AI provider payload in monitoring responses (ChatGPT-only). Enable it with
"include.rawResponse": true
to receive the full provider response alongside cloro’s structured output.
{
"prompt"
:
"What do you know about Acme Corp?"
,
"model"
:
"CHATGPT"
,
"include"
: {
"markdown"
:
true
,
"rawResponse"
:
true
}
}
Learn more in our
API Reference
.
Navigate to changelog: 16th October 2025
Added Microsoft Copilot
Microsoft Copilot (formerly Bing Chat) is now supported. You can now extract structured data from ChatGPT, Perplexity, and Copilot using the same unified interface.
To use Copilot, simply set
"model": "COPILOT"
in your API requests:
{
"prompt"
:
"What do you know about Acme Corp?"
,
"model"
:
"COPILOT"
,
"country"
:
"US"
}
All existing features work with Copilot, including:
Structured data extraction with sources & citations
Localization across different countries
Markdown format support
Citation grouping and processing
Learn more in our
API Reference
.
Navigate to changelog: 15th October 2025
Added Perplexity
Perplexity is now supported. You can now extract structured data from both ChatGPT and Perplexity models using the same unified interface.
To use Perplexity, simply set
"model": "PERPLEXITY"
in your API requests:
{
"prompt"
:
"What do you know about Acme Corp?"
,
"model"
:
"PERPLEXITY"
,
"country"
:
"US"
}
All existing features work with Perplexity, including:
Structured data extraction with sources & citations
Localization across different countries
Markdown format support
Learn more in our
API Reference
.
Navigate to changelog: 14th October 2025
Added
searchQueries
field to responses
All AI monitoring responses now include a
searchQueries
field that shows the search terms used to generate the AI response.
{
"success"
:
true
,
"result"
: {
"text"
:
"The name
\"
Acme Corporation
\"
is used in various contexts..."
,
"searchQueries"
: [
"What is Acme Corporation?"
,
"Acme Corp company overview"
,
"Acme Corporation products and services"
]
}
}
Learn more in our
API Reference
.
Navigate to changelog: 11th October 2025
Added
include.markdown
parameter
You can now receive markdown-formatted content in AI monitoring responses. Perfect for documentation workflows and content management systems.
{
"prompt"
:
"What do you know about Acme Corp?"
,
"model"
:
"CHATGPT"
,
"include"
: {
"markdown"
:
true
}
}
Learn more in our
API Reference
.
Navigate to changelog: 9th October 2025
cloro API launch
The cloro AI monitoring API is live. Extract structured data from ChatGPT, Perplexity, Microsoft Copilot, Google AI Mode and other AI models about your brand or any topic across different regions.
Key features:
AI response monitoring across different countries
Structured data extraction with sources and citations
Support for text and HTML formats
Real-time AI model tracking
Get started with our
welcome guide
or explore the
API reference
.
Was this page helpful?
Ask a question...
Navigate to changelog: 19th November 2026
Navigate to changelog: 19th November 2026
ChatGPT inline products documentation
Navigate to changelog: 19th November 2026
Navigate to changelog: 17th November 2026
include.aioverview.expand: true
Navigate to changelog: 16th November 2026
Navigate to changelog: 16th November 2026
Navigate to changelog: 12th November 2026
Navigate to changelog: 12th November 2026
POST /v1/monitor/google/product
Navigate to changelog: 9th November 2026
Navigate to changelog: 7th November 2026
Navigate to changelog: 6th November 2026
Navigate to changelog: 2nd November 2026
Why are sources and citationPills empty when web search is forced?
Navigate to changelog: 30th October 2026
Navigate to changelog: 30th October 2026
Navigate to changelog: 29th October 2026
Navigate to changelog: 27th October 2026
Navigate to changelog: 27th October 2026
Navigate to changelog: 21st October 2026
batch task creation documentation
Navigate to changelog: 19th October 2026
Navigate to changelog: 16th October 2026
Navigate to changelog: 14th October 2026
Navigate to changelog: 14th October 2026
Navigate to changelog: 12th October 2026
Navigate to changelog: 10th October 2026
Navigate to changelog: 9th October 2026
Navigate to changelog: 8th October 2026
Navigate to changelog: 8th October 2026
Navigate to changelog: 29th September 2026
Navigate to changelog: 24th September 2026
countries endpoint documentation
Navigate to changelog: 22nd September 2026
Navigate to changelog: 21st September 2026
Navigate to changelog: 18th September 2026
Navigate to changelog: 18th September 2026
Navigate to changelog: 16th September 2026
Navigate to changelog: 14th September 2026
Navigate to changelog: 9th September 2026
Navigate to changelog: 8th September 2026
Navigate to changelog: 2nd September 2026
Navigate to changelog: 31st August 2026
Navigate to changelog: 30th August 2026
Navigate to changelog: 28th August 2026
Navigate to changelog: 27th August 2026
Navigate to changelog: 26th August 2026
Navigate to changelog: 25th August 2026
Navigate to changelog: 17th August 2026
Navigate to changelog: 14th August 2026
Navigate to changelog: 11th August 2026
Navigate to changelog: 10th August 2026
Navigate to changelog: 7th August 2026
Navigate to changelog: 30th July 2026
Navigate to changelog: 27th July 2026
official AI Mode availability list
Navigate to changelog: 22nd July 2026
Navigate to changelog: 21st July 2026
Navigate to changelog: 20th July 2026
Navigate to changelog: 18th July 2026
Navigate to changelog: 18th July 2026
Navigate to changelog: 10th July 2026
Navigate to changelog: 9th July 2026
Navigate to changelog: 8th July 2026
Navigate to changelog: 7th July 2026
Navigate to changelog: 2nd July 2026
Navigate to changelog: 30th June 2026
Navigate to changelog: 30th June 2026
Navigate to changelog: 29th June 2026
Navigate to changelog: 24th June 2026
Navigate to changelog: 22nd June 2026
Navigate to changelog: 17th June 2026
Navigate to changelog: 1st June 2026
Navigate to changelog: 25th May 2026
Navigate to changelog: 22nd May 2026
Navigate to changelog: 21st May 2026
Navigate to changelog: 19th May 2026
Navigate to changelog: 19th May 2026
Navigate to changelog: 18th May 2026
Navigate to changelog: 13th May 2026
Navigate to changelog: 11th May 2026
Navigate to changelog: 10th May 2026
Navigate to changelog: 7th May 2026
Navigate to changelog: 5th May 2026
Navigate to changelog: 29th April 2026
Navigate to changelog: 25th April 2026
Navigate to changelog: 23rd April 2026
Navigate to changelog: 23rd April 2026
Navigate to changelog: 23rd April 2026
provider pricing documentation
Navigate to changelog: 22nd April 2026
Navigate to changelog: 17th April 2026
Navigate to changelog: 15th April 2026
Navigate to changelog: 9th April 2026
Navigate to changelog: 9th April 2026
Navigate to changelog: 7th April 2026
Navigate to changelog: 6th April 2026
Navigate to changelog: 1st April 2026
Navigate to changelog: 30th March 2026
product discovery capabilities
Navigate to changelog: 24th March 2026
Navigate to changelog: 20th March 2026
Navigate to changelog: 19th March 2026
Navigate to changelog: 14th March 2026
Navigate to changelog: 6th March 2026
Navigate to changelog: 3rd March 2026
Navigate to changelog: 26th February 2026
Navigate to changelog: 19th February 2026
Navigate to changelog: 10th February 2026
Navigate to changelog: 9th February 2026
Navigate to changelog: 6th February 2026
Navigate to changelog: 5th February 2026
Navigate to changelog: 3rd February 2026
Navigate to changelog: 31st January 2026
Navigate to changelog: 23rd January 2026
Navigate to changelog: 20th January 2026
Navigate to changelog: 15th January 2026
Navigate to changelog: 11th January 2026
Navigate to changelog: 18th December 2025
Navigate to changelog: 17th December 2025
Navigate to changelog: 16th December 2025
Navigate to changelog: 12th December 2025
Navigate to changelog: 23rd November 2025
Navigate to changelog: 18th November 2025
Navigate to changelog: 17th November 2025
Navigate to changelog: 15th November 2025
Navigate to changelog: 14th November 2025
Navigate to changelog: 11th November 2025
Navigate to changelog: 7th November 2025
Navigate to changelog: 6th November 2025
Navigate to changelog: 1st November 2025
Navigate to changelog: 31st October 2025
Navigate to changelog: 18th October 2025
Navigate to changelog: 17th October 2025
Navigate to changelog: 16th October 2025
Navigate to changelog: 15th October 2025
Navigate to changelog: 14th October 2025
Navigate to changelog: 11th October 2025
Navigate to changelog: 9th October 2025