Skip to content

Guides

cloro vs Bright Data for scraping AI answers at scale

Bright Data bills $1.50 per 1,000 AI answers, cloro $4.00 on its $30 Lite plan. Only cloro returns Google AI Overviews and fan-out queries; neither counts brands.

A balance scale compares a single blue circle against a tall stack of thin discs

SEO consultant, and more recently GEO, at MentionBox

On the published rates, Bright Data is the cheaper key for collecting AI answers at volume: $1.50 per 1,000 delivered records pay as you go, falling to $1.30 per 1,000 on its $499 Scale plan, against $4.00 per 1,000 ChatGPT answers on cloro's $30 Lite plan. cloro spends that difference on coverage and payload, with Google AI Overviews, fan-out queries on three engines, the entities named in the answer, US state targeting and a published rate limit, none of which appear in Bright Data's AI scraper fields. Neither one counts brands, so share of voice, position and sentiment stay in your own code. API for GEO is our editorial pick when those computed numbers are the deliverable, after hands-on testing by Mathias C.

This face-off is written for the developer who has to collect tens of thousands of AI answers a month and keep the pipeline running next quarter. Prices, endpoints, response fields and contract clauses come from each vendor's own pricing page, API reference and terms, checked on October 23, 2026. MentionBox, which publishes this site, has a commercial link with API for GEO. Five things decide the choice at volume: the shape of the bill, the engine list, the fields in the payload, the throughput one key allows, and what you are charged when a request comes back with nothing in it.

What 1,000 AI answers cost on each API

APIPublished price per 1,000 AI answersEntryHow one answer is billed
API for GEO$13, down to $7 at volumepay as you go, no minimum publishedone credit per prompt, engine and locale, metrics included
cloro$4.00 on Lite, $2.00 on Hobby, $1.55 at the top tier$30 a month for 37,500 credits; free plan of 5004 or 5 credits per answer by engine, plus add-ons
Bright Data AI Scrapers$1.50 pay as you go, $1.30 on Scalenone; 5,000 free credits a month, no cardone credit per delivered record, same price on every engine

Published rates from each vendor's pricing page, checked on October 23, 2026. For API for GEO, the engine coverage beyond the six it lists, along with its webhooks, OpenAPI spec and official SDKs, was confirmed in our hands-on test on October 26, 2026 and is not yet documented on apiforgeo.com.

Bright Data sells records. The Web Scraper API pricing page publishes $1.50 per 1,000 records pay as you go and a $499 Scale plan that includes 384,000 records, then $1.30 per 1,000 beyond it, with no upfront commitment. Its FAQ states the rule plainly: one credit per record, billed "per delivered record". A ChatGPT answer and a Google AI Mode answer cost the same, which makes the monthly forecast a single multiplication. New accounts get 5,000 free credits a month without a card, shared since June 2026 across Web Unlocker, the SERP API, the Web Scraper API and the MCP server.

cloro sells credits inside a monthly plan, and each engine has its own credit cost. The ladder on its pricing page runs from a free plan of 500 credits with one concurrent job, through Lite at $30 for 37,500 credits ($0.80 per 1,000 credits) and Hobby at $100 for 250,000 ($0.40), up to $5,000 for 16,391,783 credits ($0.31). The providers guide prices the engines:

Engine on cloroCredits per async requestPer 1,000 answers on LitePer 1,000 answers on Hobby
ChatGPT with web search5$4.00$2.00
Copilot5$4.00$2.00
Google AI Overview5$4.00$2.00
Perplexity4$3.20$1.60
Gemini4$3.20$1.60
Google AI Mode4$3.20$1.60

Credit costs from the cloro providers guide, plan credit prices from its pricing page, both checked on October 23, 2026. Our arithmetic multiplies the credits per answer by the plan's price per 1,000 credits.

Four surcharges move those numbers, all documented and all easy to switch on by accident: a synchronous call on /v1/monitor/* adds 2 credits where the async path adds none, US state targeting adds 2 on ChatGPT, Copilot, Perplexity and Gemini, the ChatGPT add-ons for fan-out, shopping, raw response and ads add 2 each, and a follow-up turn in a ChatGPT conversation has cost 1 extra credit since October 14, 2026. The pricing page also prints a 7-credit line for a ChatGPT "full response" next to the 5-credit web search row, so settle which figure applies to your payload before you budget a month of it.

The bill at 15,000 and at 150,000 answers a month

Take 100 tracked prompts on five engines, collected daily. That is 15,000 AI answers a month on either vendor, and the same basket costs 22 cloro credits per prompt per day: ChatGPT 5, Copilot 5, Perplexity 4, Gemini 4, Google AI Mode 4.

Route15,000 answers a monthPer 1,000150,000 answers a monthPer 1,000
API for GEO, pay as you go$195.00$13.00$1,050 to $1,950 by tier$7.00 to $13.00
cloro, smallest plan that fits$100.00, Hobby$6.67$500.00, Growth$3.33
Bright Data, pay as you go$22.50$1.50$225.00$1.50

Our arithmetic on the rates each vendor published, checked on October 23, 2026, on one stated basket: 100 or 1,000 prompts, five engines, one run a day, async requests, no add-ons. The two runs consume 66,000 and 660,000 cloro credits. Change the engine mix or the plan and every figure moves.

Four details inside that table decide more than the sticker price.

  • Plan fill. The 15,000-answer run burns 66,000 cloro credits, overflowing Lite's 37,500 and using a quarter of Hobby's 250,000. You pay $100 either way, so the effective rate is $6.67 per 1,000 rather than the $2.00 the credit price suggests. Filling the plan is what brings cloro near Bright Data's rate.
  • The tier gap. The 150,000-answer run needs 660,000 credits, and Starter at $250 buys 650,000. Being 10,000 credits short costs $250, because standard plans stop with a 403 INSUFFICIENT_CREDITS rather than billing an overage, a behaviour cloro documents in its billing guide. Only annual Enterprise contracts allow overage.
  • The floor in a quiet month. cloro's plan fee is due whether you call the API or not, so a pause in collection still costs $100 or $500 in the two rows above. Bright Data pay as you go and API for GEO both fall to zero.
  • Where Scale pays off. Bright Data's $499 plan equals pay as you go at 332,667 records a month, which is $499 divided by $1.50 per 1,000. Below that volume the plan costs more than the meter.

Which engines each key reaches

Engine, of the 10 we trackAPI for GEOcloroBright Data
ChatGPTyesyes, 5 creditsyes
Perplexityyesyes, 4 creditsyes
Google AI Modeyesyes, 4 creditsyes
Copilotyesyes, 5 creditsyes
Geminiyes, from our testyes, 4 creditsyes, European countries not available
Google AI Overviewsyes, from our testyes, 5 credits, as an add-on to Google Searchno, served by the separate SERP API
Grokyesdiscontinued on October 15, 2026listed, currently unavailable
Claudeyesnono
DeepSeekyes, from our testnono
Meta AIyes, from our testnono
Total you can call today1065

Engine lists published in each vendor's documentation, checked on October 23, 2026. For API for GEO, ChatGPT, Perplexity, Google AI Mode, Claude, Copilot and Grok are listed on apiforgeo.com, and Gemini, Google AI Overviews, DeepSeek and Meta AI were confirmed in our hands-on test on October 26, 2026.

Three gaps in that grid are worth checking before you promise a dashboard column. Google AI Overviews sit on different products. cloro returns the overview inside a Google Search request through include.aioverview, with the text, the sources, the citation pills, the related links Google groups without citing, the videos, the places and the ads in the block, all documented on its AI Overview reference. Bright Data's AI visibility page serves Google AI Overview from the SERP API instead, so a second product, a second rate card and a second integration. If Google is the surface you care about most, our face-off on SerpApi and DataForSEO for Google AI Overviews prices the specialists on the same basis.

Grok has gone on both. cloro's providers guide marks it "Discontinued on 15 Oct 2026", and Bright Data's release notes record its Grok scraper unavailable in May 2026, back on June 11, unavailable again on June 14, and dropped from version 3.0.0 of its MCP server on October 11, 2026 because the underlying scrapers "no longer exist".

Claude is absent from both, and cloro says why in its provider documentation: access "conflicts with Anthropic's terms of service". An interface scraper reaches what a logged-out browser reaches, so an engine behind an account stays out of range, which is also where DeepSeek and Meta AI sit. Our ranking of AI answer scraping APIs scores fourteen of these keys on the same four criteria.

What comes back for one prompt

FieldAPI for GEOcloroBright Data
Answer textyestext, markdown, html, rawResponseanswer_text, answer_html
Cited sourcesranked and sorted by categorysources: position, url, label, description, footnote, datePublishedcitations: url, title, description, icon
Search results behind itnot publishedcitationPills with domain and positionsearch_sources: url, title, snippet, rank
Brands named in the answerentities[], aliases collapsedentities: type, name, productnot published
Fan-out queriesown endpointssearchQueries on ChatGPT, Perplexity and Copilotnot published
ShoppingyesshoppingCards, inlineProducts, product detailsproduct cards on ChatGPT, shopping flag on Perplexity
Ads, maps and places in the answernot publishedads, map and places with rating and addressis_map flag
Whether live search rannot publisheddisableWebSearch on the requestweb_search_triggered
Collection timestampnot publishedcreatedAt and latencyMs on the taskprompt_sent_at
Brand position, share of voice, sentimentyesnono
Themes and per-brand perceptionyesnono
Countries187 on its published list250, plus US states on four enginesper-engine lists in a public repository

Field names published in each vendor's API reference, checked on October 23, 2026.

The shared floor is the answer text and the cited URLs, which is what keeps both prices near or below $4 per 1,000. Above it, cloro's payload is the wider one: the ChatGPT endpoint returns the entities named in the answer, the ad units rendered inside it, the map cards for local prompts and the searchQueries array, the sub-questions the engine fired before writing. Bright Data describes "up to 25 metadata fields" per LLM record, and its ChatGPT record carries the two worth having for an audit: web_search_triggered, which says whether the engine actually searched, and prompt_sent_at, the moment the scraper pressed send.

Fan-out is the field we would pay the extra 2 credits for. In an analysis we ran in October 2026, the prompts that exposed their sub-queries showed a median of four distinct fan-out queries each, so an answer logged without them hides most of what the engine searched. Our explainer on query fan-out in AI search covers how those sub-searches decide which pages get cited, and tracking a brand in ChatGPT with an API sets out the counting code both of these APIs leave to you, formula by formula.

How to send one prompt on each API

cloro is a task API. One POST creates one async task, and the same payload works on the synchronous path for 2 extra credits:

curl -X POST "https://api.cloro.dev/v1/async/task" \
  -H "Authorization: Bearer $CLORO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "taskType": "CHATGPT",
    "idempotencyKey": "chatgpt-best-crm-2026-11-02",
    "webhook": { "url": "https://your-app.example/cloro-hook" },
    "payload": {
      "prompt": "best crm for small business",
      "country": "US",
      "include": { "markdown": true, "searchQueries": true }
    }
  }'
{
  "success": true,
  "task": {
    "id": "b27a21e1-7c39-4aa2-a347-23e828c426f9",
    "taskType": "CHATGPT",
    "status": "QUEUED",
    "priority": 5,
    "createdAt": "2026-11-02T15:00:00.000Z",
    "latencyMs": null
  },
  "credits": { "creditsToCharge": 7, "creditsCharged": null }
}

The response states the price before the work happens: 5 credits for the ChatGPT base request and 2 for the fan-out add-on, with creditsCharged filled in once the task reaches COMPLETED. Use idempotencyKey on a nightly job, since a replayed batch returns RESOURCE_ALREADY_EXISTS instead of a second charge. Batches go to POST /v1/async/task/batch, up to 500 tasks per request, with a per-task result array so one bad task does not cost you the batch.

Bright Data is a dataset API. One POST triggers a collection against a dataset ID and returns a snapshot to download later:

curl -X POST "https://api.brightdata.com/datasets/v3/trigger\
?dataset_id=gd_m7aof0k82r803d5bjm&include_errors=true" \
  -H "Authorization: Bearer $BRIGHTDATA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input":[
    {"prompt":"best crm for small business","country":"US","web_search":true},
    {"prompt":"crm pricing for a 20 person team","country":"US","web_search":true}
  ]}'
curl "https://api.brightdata.com/datasets/v3/snapshot/sd_m1a2b3c4d5e6f7g8h?format=json" \
  -H "Authorization: Bearer $BRIGHTDATA_API_KEY"

The dataset ID for ChatGPT Search is gd_m7aof0k82r803d5bjm, published in its ChatGPT scraper documentation, and the async guide tells you to poll the progress endpoint "until it shows ready, and stop if it shows failed or canceled". The synchronous twin, POST /scrape, returns the data inline when collection finishes inside a minute and otherwise answers HTTP 202 with a snapshot_id, so one client has to handle both outcomes.

How much each key collects per hour

Throughput is the number to check before you promise a client a report by 9 a.m., and the two vendors publish different halves of it.

  • cloro publishes both ceilings. Its concurrency guide gives 1,000 requests per second per endpoint per API key, and the pricing page sets concurrent jobs by plan: 1 on Free, 10 on Lite, 20 on Hobby, 50 on Starter, 75 on Growth and up to 255 at the top. Past the limit you get a 429 immediately rather than a queue, and the guide tells you to "back off exponentially rather than re-sending at once". With the 30 to 45 second average its home page advertises, Hobby's twenty slots carry roughly 1,600 to 2,400 answers an hour, so the 15,000-answer month above is about seven hours of wall clock.
  • Bright Data publishes volume instead of rate. No requests-per-second figure appears in its documentation. It sizes the job: up to 5,000 inputs in one call, up to 1 GB of input per async job, and a snapshot downloadable for 30 days.
  • Retention decides your plumbing. cloro keeps a task result readable through the REST API for 24 hours, so your own database has to be the store of record the same day. Bright Data's 30-day snapshots, plus delivery to Amazon S3, Google Cloud Storage, Pub/Sub, Azure Storage, Snowflake and SFTP in JSON, NDJSON, CSV, Parquet or gzip, let the vendor hold the raw layer while you load the warehouse. cloro signs its webhook with HMAC-SHA256 in an X-Cloro-Signature header and retries five times; Bright Data splits the callback in two URLs, notify for the status and endpoint for the records.

Reliability is the one place where neither vendor gives you much to work with. Bright Data's ChatGPT page advertises a "99.2% Avg. success rate", cloro's home page claims "99.99% uptime", and neither publishes the method behind the figure. Both also publish comparison pages that measure the other, which is a vendor testing a rival. In our view the only number worth budgeting against is your own: run the same 200 prompts through both keys for a week and count, per engine, the share of requests that came back with a non-empty answer and at least one source.

What you pay for when the answer comes back empty

At 15,000 answers a month, a 5% empty rate is 750 requests, and the two rate cards treat them differently.

cloro charges on success only. Its billing guide states that it charges "a sync request when it returns 200 (success: true), an async task when it reaches COMPLETED", and the pricing page repeats that failed extractions and system errors consume no credits. Two caveats cost real money. An optional block that comes back empty inside a successful response counts as a valid result and is still billed, so a shopping add-on on a prompt with no product cards is 2 credits spent. And when Google shows no AI Overview, including in a country where Google does not offer one, the AI Overview reference says the response is a 200 with no overview key and the add-on is still charged. Credits "expire at the end of each billing cycle and do not roll over", and the pricing page states "We do not offer refunds".

Bright Data charges on delivery, with its own caveat. The pricing page promises "no charges for failed deliveries", and the scraper FAQ narrows it: records that fail because the input was incorrect are still billed, "because the request consumed resources". Validate your prompt list and your country codes before a 5,000-input job rather than after. The 5,000 free monthly credits renew on the first and do not accumulate.

The two changelogs show the same pressure arriving in two forms. Bright Data's release notes record its ChatGPT scraper unavailable twice in June 2026 and its Grok scraper retired for good. cloro's changelog records price rises instead: Copilot to 5 credits and a +2 surcharge on every sync request on April 23, 2026, then Perplexity to 4 credits on August 25, 2026 after what it describes as a 70% cost increase on its side. An engine tightening its anti-bot defences reaches you either as an outage or as a new rate card.

What the two contracts allow

Both vendors anchor the practice in the same claim, that they only read what an anonymous visitor can read. Bright Data states that "all data is collected exclusively from publicly accessible, unauthenticated web interfaces", and cloro's ChatGPT reference states that "cloro does not call the ChatGPT API: each request runs a real browser session against ChatGPT's web interface", logged out and routed through a proxy in the country you ask for. Bright Data also argues for the method, writing that "vendor APIs return sanitized 'API-mode' responses that differ from what real users experience". Our comparison of the OpenAI web search tool and ChatGPT answers tests that claim against the official route, field by field.

Resale is where both say no.

  • cloro, at 2.3(a) of its terms, forbids you to "sell, resell, license, sublicense, distribute, rent, or lease the Services, or make the Services available to any third party". Section 6.2 leaves you "all rights, title, and interest in and to your data and any content that you submit or retrieve", so the clause targets access to the service rather than the records you store. No white label clause exists either way, and section 13.1 puts the contract under the laws of Ireland.
  • Bright Data, in its service agreement last updated October 12, 2026, requires "Bright Data's prior written authorization" to resell the Service at 3.8.III, and forbids you to "distribute, transmit, reproduce, publish, license, transfer, or sell any Data in order to offer a similar or competitive product" at 3.9.I. Resale runs through a partner agreement. Against that, it publishes the compliance file a procurement team asks for: ISO 27001, SOC 2 Type II, CSA STAR Level 1, and GDPR and CCPA statements. The contracting entity is Bright Data Ltd. in Israel, or Bright Data, Inc. for clients in the United States.
  • API for GEO reads the other way, naming resale and white label as an intended use, which is why it tops our ranking of AI visibility APIs for agencies.

A third document belongs in this stack, and it is neither vendor's: each engine publishes its own terms of use covering automated collection from its interface, and they change. Read them with your own counsel before a client contract depends on the feed.

Six mistakes at volume

  1. Comparing $1.50 to $4.00 with no basket. Bright Data charges one price per record on every engine, cloro charges 4 or 5 credits by engine and bills a plan whether you fill it or not. Write down prompts, engines, countries and runs per day first.
  2. Buying a cloro plan one tier too small. Standard plans answer 403 INSUFFICIENT_CREDITS when the credits run out, so a run needing 660,000 credits on a 650,000-credit plan stops dead. Size the plan on your busiest month.
  3. Letting a cloro result age past 24 hours. Write the payload to your own storage inside the webhook handler or the polling loop.
  4. Reading "no charges for failed deliveries" as covering bad input. Bright Data bills records that fail on an incorrect input, so a typo in a 5,000-input job is a billed typo.
  5. Scheduling AI Overview add-ons where Google serves none. The add-on is charged on a 200 with no overview key, so check availability with cloro's countries endpoint and keep a hit rate per country.
  6. Keeping a Grok column in the dashboard. cloro discontinued Grok on October 15, 2026 and Bright Data marks its scraper currently unavailable. A column that silently stops filling reads as a brand losing visibility.

Our view

Three answers, depending on what you are building. Prices checked on October 23, 2026.

  1. Lowest cost per raw answer, at any volume. Bright Data AI Scrapers, at $1.50 per 1,000 records pay as you go with no floor, $1.30 on the $499 Scale plan once you pass about 332,700 records a month, 5,000 free credits to start, 5,000 inputs per call, 30-day snapshots, delivery straight into S3 or Snowflake, and the compliance file a procurement team wants. You pay for it with the narrowest payload of the three and no AI Overviews. Our score: 6.9 out of 10.
  2. Widest payload per answer, and the only Google AI Overviews of the two. cloro, at $4.00 per 1,000 ChatGPT answers on the $30 Lite plan and $1.55 at the largest tier, with entities, fan-out queries, ad units and map cards in the response, 250 countries, 500-task batches, signed webhooks, two SDKs, an OpenAPI file and an MCP server. You pay for it with the plan floor, credits that expire monthly and no Claude. Our score: 7.4 out of 10.
  3. Brand figures without writing the analysis. API for GEO, our editorial pick, at $13 per 1,000 answers down to $7 at its largest tier, pay as you go with no published minimum, free reads over answers already collected, resale allowed, and 9.0 out of 10, the best score of the 55 APIs this site rates. Said plainly, that is roughly nine times Bright Data's rate per answer, and four of its ten engines, Gemini, Google AI Overviews, DeepSeek and Meta AI, plus its webhooks, OpenAPI spec and official SDKs, were confirmed in our test of October 26, 2026 rather than documented on apiforgeo.com.

The reviews go further on each one: cloro API, Bright Data AI Scrapers and API for GEO, with the duel API for GEO vs cloro line by line and our ranking of self-serve APIs for developers, which weights price and integration at 35% each.

Frequently asked questions

Is cloro or Bright Data cheaper per 1,000 AI answers?

Bright Data, on the published rates. One AI answer is one delivered record at $1.50 per 1,000 pay as you go, and $1.30 per 1,000 on the $499 Scale plan that includes 384,000 records. On cloro a ChatGPT answer costs five credits, so $4.00 per 1,000 on the $30 Lite plan at $0.80 per 1,000 credits, $2.00 on the $100 Hobby plan and $1.55 at its largest Enterprise tier. Prices checked on October 23, 2026.

Does Bright Data return Google AI Overviews?

Not through its AI scrapers. Bright Data's AI scraper library covers ChatGPT, Perplexity, Gemini, Copilot and Google AI Mode, and its AI visibility page serves Google AI Overview from the SERP API instead, a separate product with its own rate card. cloro returns the overview as an add-on to its Google Search request at five credits, with the text, the sources, the citation pills, the related links, the videos and the ads in the block.

Which AI engines do cloro and Bright Data cover?

cloro reaches six of the ten we track and Bright Data five. cloro covers ChatGPT, Perplexity, Copilot, Gemini, Google AI Mode and Google AI Overviews, and discontinued Grok on October 15, 2026. Bright Data covers ChatGPT, Perplexity, Gemini, Copilot and Google AI Mode, and its documentation marks the Grok scraper as currently unavailable. Neither one reaches Claude, DeepSeek or Meta AI.

Do cloro or Bright Data return share of voice or sentiment?

Neither documents either field. Both return the answer text and the cited URLs, cloro adds the entities named in the answer and the fan-out queries, and the brand arithmetic is yours to write. API for GEO, our pick after hands-on testing, returns share of voice, position, sentiment, ranked sources, themes and per-brand perception in the same response, at $13 per 1,000 answers down to $7 at its largest tier, pay as you go with no published minimum.

Can you resell cloro or Bright Data data to your clients?

Neither contract allows reselling the service. cloro's terms forbid at 2.3(a) any attempt to "sell, resell, license, sublicense, distribute, rent, or lease the Services", while 6.2 leaves you the rights in the data you retrieve. Bright Data's service agreement, last updated October 12, 2026, requires "Bright Data's prior written authorization" to resell the Service and bars distributing Data "in order to offer a similar or competitive product".

What happens to the bill when an AI answer comes back empty?

cloro charges on success only, meaning a sync request that returns 200 or an async task that reaches COMPLETED, and its credits never roll over to the next month. One exception costs money at scale: when Google shows no AI Overview the response is a 200 with no overview key, and the add-on is still charged. Bright Data states it makes "no charges for failed deliveries", and its FAQ adds that records failing because the input was incorrect are still billed.