>which sources does chatgpt cite for "best crm for a small team"?cloro - scrape_chatgpt (MCP)(prompt: "best crm for a small team",country: "US"){"text": "For a small team, the strongest options…","sources": [… +41 lines (ctrl+o to expand)ChatGPT cites 3 sources for that prompt:1. wikipedia.org/wiki/CRM2. reddit.com/r/sales3. youtube.com/watch
Structured data with a single API
Covers the main SEO and AI SEO providers, at any scale and across regions. Every engine takes the same country targeting, so one run repeats in the next market.
Direct provider APIs return the answer text, not its sources. cloro parses sources and citations from every response, so you know which pages to work on.
Each provider requires separate API integration, creating technical overhead. cloro puts every engine behind one API and one response shape, so a new engine is a config change.
1,000 prompts per engineCreditsChatGPT (web search)5 credits × 1,0005,000AI Mode4 credits × 1,0004,000Gemini4 credits × 1,0004,000Cost of the runKnown before it starts5,000 credits9,000 credits13,000 credits
Know the cost up front
Token-based pricing varies by model and provider, making costs unpredictable. cloro charges a published credit cost per request, so you can price a run in advance.
Search API accuracy against cost
We ran three public question sets through nine search APIs: one search per question, and the same model answers from each set of results. Cost is the list price per 1,000 searches on the plan that covers 100,000 searches a month.
SimpleQA Verified, 1,000 questions. Short factual questions with one correct answer. Perplexity, cloro, Parallel, Keenable and Exa tie within the margin of error, and cloro costs less than half as much. cloro beat SerpApi, Brave, You.com, Tavily and Firecrawl.
↑ Accuracy on SimpleQA Verified
86%90%94%98%
↖ More accurate, lower cost
cloro
cloro
Accuracy on SimpleQA Verified: 96.4%
Cost per 1,000 searches: $1.92
95% interval: 95.1% to 97.4%
Starter, $250 a month for 650,000 credits, 5 credits a search
Perplexity
Perplexity
Accuracy on SimpleQA Verified: 96.9%
Cost per 1,000 searches: $5
95% interval: 95.6% to 97.8%
Search API, $5 per 1,000 requests
Parallel
Parallel
Accuracy on SimpleQA Verified: 96.2%
Cost per 1,000 searches: $5
95% interval: 94.8% to 97.2%
Search API, pay as you go, $5 per 1,000 requests
Keenable
Keenable
Accuracy on SimpleQA Verified: 96%
Cost per 1,000 searches: $4
95% interval: 94.6% to 97%
Pay as you go, $4 per 1,000 requests after 100,000 free requests a month
Exa
Exa
Accuracy on SimpleQA Verified: 95.8%
Cost per 1,000 searches: $7
95% interval: 94.4% to 96.9%
Search (auto), pay as you go, $7 per 1,000 requests
SerpApi
SerpApi
Accuracy on SimpleQA Verified: 93.9%
Cost per 1,000 searches: $7.25
95% interval: 92.2% to 95.2%
Searcher, $725 a month for 100,000 searches
Brave
Brave
Accuracy on SimpleQA Verified: 90.5%
Cost per 1,000 searches: $5
95% interval: 88.5% to 92.2%
Search, $5 per 1,000 requests
You.com
You.com
Accuracy on SimpleQA Verified: 89.2%
Cost per 1,000 searches: $5
95% interval: 87.1% to 91%
Web Search API, pay as you go, $5 per 1,000 calls
Tavily
Tavily
Accuracy on SimpleQA Verified: 89.2%
Cost per 1,000 searches: $5
95% interval: 87.1% to 91%
Growth, $500 a month for 100,000 basic searches
Firecrawl
Firecrawl
Accuracy on SimpleQA Verified: 88.8%
Cost per 1,000 searches: $3.49
95% interval: 86.7% to 90.6%
Standard, $99 a month for 100,000 credits plus 100,000 extra credits at $5 per 2,000; 2 credits a search
$0$2$4$6$8
Cost per 1,000 searches →
SimpleQA Verified, 1,000 questions, run in November 2026. One search per question, the question text as the query (GitHub). The same reader, gemini-3.8-flash, answers only from each set of results, and gemini-pro-latest grades each answer with the simple-evals grader prompt. List price per 1,000 searches on each provider's cheapest self-serve plan that covers 100,000 searches a month.
FreshQA, 358 questions. Questions whose answer can change over time, a third of them with a false premise. cloro scores highest, within the margin of error of SerpApi, Exa, Parallel, Perplexity and Brave. cloro answered more questions correctly than You.com, Keenable, Tavily and Firecrawl.
↑ Accuracy on FreshQA
78%82%86%90%
↖ More accurate, lower cost
cloro
cloro
Accuracy on FreshQA: 88.5%
Cost per 1,000 searches: $1.92
95% interval: 84.8% to 91.4%
Starter, $250 a month for 650,000 credits, 5 credits a search
SerpApi
SerpApi
Accuracy on FreshQA: 86.9%
Cost per 1,000 searches: $7.25
95% interval: 83% to 90%
Searcher, $725 a month for 100,000 searches
Exa
Exa
Accuracy on FreshQA: 86.6%
Cost per 1,000 searches: $7
95% interval: 82.7% to 89.7%
Search (auto), pay as you go, $7 per 1,000 requests
Parallel
Parallel
Accuracy on FreshQA: 86%
Cost per 1,000 searches: $5
95% interval: 82.1% to 89.2%
Search API, pay as you go, $5 per 1,000 requests
Perplexity
Perplexity
Accuracy on FreshQA: 86%
Cost per 1,000 searches: $5
95% interval: 82.1% to 89.2%
Search API, $5 per 1,000 requests
Brave
Brave
Accuracy on FreshQA: 85.8%
Cost per 1,000 searches: $5
95% interval: 81.8% to 89%
Search, $5 per 1,000 requests
You.com
You.com
Accuracy on FreshQA: 84.9%
Cost per 1,000 searches: $5
95% interval: 80.8% to 88.3%
Web Search API, pay as you go, $5 per 1,000 calls
Keenable
Keenable
Accuracy on FreshQA: 84.4%
Cost per 1,000 searches: $4
95% interval: 80.2% to 87.8%
Pay as you go, $4 per 1,000 requests after 100,000 free requests a month
Tavily
Tavily
Accuracy on FreshQA: 82.7%
Cost per 1,000 searches: $5
95% interval: 78.4% to 86.2%
Growth, $500 a month for 100,000 basic searches
Firecrawl
Firecrawl
Accuracy on FreshQA: 80.2%
Cost per 1,000 searches: $3.49
95% interval: 75.7% to 84%
Standard, $99 a month for 100,000 credits plus 100,000 extra credits at $5 per 2,000; 2 credits a search
$0$2$4$6$8
Cost per 1,000 searches →
FreshQA, 358 questions, run in November 2026. One search per question, the question text as the query (GitHub). The same reader, gemini-3.8-flash, answers only from each set of results, and gemini-pro-latest grades each answer with the simple-evals grader prompt. List price per 1,000 searches on each provider's cheapest self-serve plan that covers 100,000 searches a month. FreshQA version of April 21, 2026, test split, without the fast-changing questions. Some answer keys may be out of date.
FRAMES, 824 questions. Multi-hop questions that need facts from two or more Wikipedia pages. cloro and SerpApi score highest, 27 to 64 points ahead of the other eight. With one search, those eight could not answer 372 to 673 of the 824 questions.
↑ Accuracy on FRAMES
10%30%50%70%90%
↖ More accurate, lower cost
cloro
cloro
Accuracy on FRAMES: 79.9%
Cost per 1,000 searches: $1.92
95% interval: 77% to 82.5%
Starter, $250 a month for 650,000 credits, 5 credits a search
SerpApi
SerpApi
Accuracy on FRAMES: 76.8%
Cost per 1,000 searches: $7.25
95% interval: 73.8% to 79.6%
Searcher, $725 a month for 100,000 searches
Exa
Exa
Accuracy on FRAMES: 49.9%
Cost per 1,000 searches: $7
95% interval: 46.5% to 53.3%
Search (auto), pay as you go, $7 per 1,000 requests
Parallel
Parallel
Accuracy on FRAMES: 45.6%
Cost per 1,000 searches: $5
95% interval: 42.3% to 49%
Search API, pay as you go, $5 per 1,000 requests
Perplexity
Perplexity
Accuracy on FRAMES: 44.2%
Cost per 1,000 searches: $5
95% interval: 40.8% to 47.6%
Search API, $5 per 1,000 requests
Keenable
Keenable
Accuracy on FRAMES: 34.5%
Cost per 1,000 searches: $4
95% interval: 31.3% to 37.8%
Pay as you go, $4 per 1,000 requests after 100,000 free requests a month
Brave
Brave
Accuracy on FRAMES: 25.8%
Cost per 1,000 searches: $5
95% interval: 23% to 28.9%
Search, $5 per 1,000 requests
You.com
You.com
Accuracy on FRAMES: 22.1%
Cost per 1,000 searches: $5
95% interval: 19.4% to 25%
Web Search API, pay as you go, $5 per 1,000 calls
Tavily
Tavily
Accuracy on FRAMES: 18.1%
Cost per 1,000 searches: $5
95% interval: 15.6% to 20.9%
Growth, $500 a month for 100,000 basic searches
Firecrawl
Firecrawl
Accuracy on FRAMES: 15.8%
Cost per 1,000 searches: $3.49
95% interval: 13.4% to 18.4%
Standard, $99 a month for 100,000 credits plus 100,000 extra credits at $5 per 2,000; 2 credits a search
$0$2$4$6$8
Cost per 1,000 searches →
FRAMES, 824 questions, run in November 2026. One search per question, the question text as the query (GitHub). The same reader, gemini-3.8-flash, answers only from each set of results, and gemini-pro-latest grades each answer with the simple-evals grader prompt. List price per 1,000 searches on each provider's cheapest self-serve plan that covers 100,000 searches a month. FRAMES is usually run with an agent that searches many times. Here each API got one search, so the score shows how much of the answer one results page holds. You.com rejects queries longer than 50 words, so its 42 longest questions count as not attempted.
How the cloro integration works
Choose your platform, send a query, get structured data back: markdown, text, or HTML, plus parsed sources, citations, query fan-out, and shopping blocks.
Request example
python
from cloro import Cloro
client = Cloro(api_key="sk_live_your_api_key_here")
response = client.monitor.chatgpt(
prompt="What do you know about Tesla's latest updates?",
country="US",
state="CA",
include={"markdown": True},
)
print(response["result"]["text"])
typescript
import { Cloro } from "@cloro-dev/cloro";
const client = new Cloro({ apiKey: "sk_live_your_api_key_here" });
const response = await client.monitor.chatgpt({
prompt: "What do you know about Tesla's latest updates?",
country: "US",
state: "CA",
include: { markdown: true },
});
console.log(response.result.text);
curl
curl -X POST https://api.cloro.dev/v1/monitor/chatgpt \
-H "Authorization: Bearer sk_live_your_api_key_here" \
-H "Content-Type: application/json" \
-d '{
"prompt": "What do you know about Tesla'\''s latest updates?",
"country": "US",
"state": "CA",
"include": {
"markdown": true
}
}'
Engineers, founders and marketers who replaced their own scrapers with one API.
4.6/5 from 99 reviews on
It’s fast, reliable, and easy to use and integrate. cloro delivers dependable answers that match what real users see on the web interfaces of ChatGPT, Gemini, Perplexity, and Copilot, all through a single API, which is very useful for analyzing results.
AB
akhil b.
I like cloro's great API which is fast, reliable, and up-to-date. Additionally, I appreciate the nice pricing and good developer user experience. The initial setup was super easy, which made getting started a breeze.
JB
Julian B.
Founder, Framebird.io
What I like best about cloro is that it shows the actual AI response for a specific query, including which brands were mentioned, which competitors appeared, and what sources were cited. It gives us clear, query-level visibility rather than just high-level traffic data.
BA
Bindu A.
Applied AI Engineer
I appreciate cloro for its cost and speed. With our large volume of searches, the cost is really important for our unit economics. Setting it up was very easy.
AG
Ahmed G.
Small-Business
I like that cloro allows me to use just one API to handle different calls for various AI models, which makes it very convenient. I also find the playground set up on their website extremely useful because I can test API calls immediately without setting up my own environment.
JR
John Rey B.
Full-stack Digital Marketer
I use cloro for my startup rekkd.io, which helps restaurant owners and marketers see how different LLMs perceive their business. I really like cloro's usability, and I appreciate having full information about sources and visited resources when I send a prompt.
AM
Alexander M.
Lead Solutions Architect
Querying ChatGPT, Gemini, Perplexity, Copilot, and Google AI as part of our brand analytics pipeline used to mean maintaining separate SDKs and authentication flows for each, but with cloro, it's just one POST request with the same payload structure regardless of the platform. This alone saved us days of integration work.
Verified User in Program Development
Small-Business
cloro excels in high-fidelity citation mapping, reliably extracting exact source URLs and citations into a clean, structured JSON format, making it robust for verifying AI claims and tracking brand visibility. The setup was smooth and took under 15 minutes.
YS
you s.
Small-Business
The JSON response is well-structured and parsed — text, citations, sources, all separated cleanly. No need to parse raw HTML like with other providers. Response time around 20 seconds for ChatGPT. Simple REST API, easy to integrate.
Our API delivers structured responses in 30-45 seconds on average. We handle all the complexity of automation and parsing to give you clean markdown content and citations in near real-time.
What data do you extract from AI models?
We extract the complete AI response in parsed markdown format, all source & citations with URLs and labels. Everything is delivered as structured JSON for easy integration into your systems.
Which models do you support?
We support ChatGPT (with Web Search), Google Search, Google News, Google AI Overview, Google AI Mode, Google Gemini, Bing Search, and Bing AI Mode. Our unified API lets you extract data from all these sources using a consistent interface.
Can I see the kind of data you return before signing up?
Yes. The AI Visibility Leaderboard is built on this API and republished every Monday: which brands AI engines such as ChatGPT, Google AI Mode and Gemini name and cite across twelve software categories, alongside every domain they cited as a source. No signup, and each weekly snapshot stays available so you can cite a specific week.
Do you support asynchronous requests?
Yes! We offer an async endpoint that allows you to submit tasks and receive the results via webhook. This is ideal for high-volume scraping and ensures reliability without keeping connections open.
How does concurrency work?
Concurrency limits depend on your subscription plan. For example, business plans often start with higher concurrency limits (e.g., 100). If you exceed your limit, you will receive a 429 status code. We recommend implementing client-side queuing or using our async endpoint.
Can I increase my concurrency limits?
You can increase your concurrency by subscribing to a higher-volume package. Since we dedicate resources exclusively to your organization, we want to ensure those resources are fully utilized and not just spiky traffic. However, if you are indeed limited by concurrency (constantly using your concurrency more than 70% of the time), you should contact us so we can adjust.
Can I get the raw HTML?
Yes. By default, we return parsed structured JSON to keep responses lightweight. However, you can retrieve the full raw HTML of the result page by setting include.html: true in your request.
Can I extract data from specific geographic regions?
Yes, we support all countries globally, plus US state-level targeting on ChatGPT, Gemini and Bing AI Mode. Pass the state parameter (e.g. CA) alongside country: US and we route the request through a proxy in that state. See the states endpoint for the full list. Note that some models (like Gemini) may have regional availability restrictions (e.g., EU).
What happens if a request times out?
You generally don't need to implement complex timeout logic. Our system automatically retries requests up to 10 times with a maximum duration of 5 minutes to ensure high success rates.
When are credits deducted?
Credits are deducted only when we successfully extract and deliver AI response data. Failed extractions or system errors do not consume your credits.
Do credits roll over from month to month?
No, credits do not roll over from month to month. Each billing cycle, you receive your allocated credits based on your subscription plan. Any unused credits expire at the end of the billing period.
Run SEO and AI SEO scrapes at scale
Track your brand and your competitors across global SERPs, with low latency.