cloro
Sign in Get an API key
ChatGPT · engine endpoint

Turn ChatGPT answers into rows you can query.

Send a prompt, get back the exact answer ChatGPT gave, the sources it cited and every brand it named — as structured JSON or clean Markdown. No headless browser, no scraping stack, no reverse-engineering the app.

Scrape your first prompt Read the docs
● Answer + citations + brands ● Batch & region-aware ● Pay per successful call
POST /v1/chatgpt
# prompt ChatGPT, get structured data back
curl https://api.cloro.cloud/v1/chatgpt \
  -H "Authorization: Bearer $CLORO_KEY" \
  -d '{ "prompt": "what CRM do you recommend for a small B2B team?",
        "country": "us",
        "format": "json" }'

# → response
{
  "engine": "chatgpt",
  "prompt": "what CRM do you recommend...",
  "answer": "For a small B2B team, a few solid options...",
  "sources": [
    { "rank": 1, "domain": "g2.com" },
    { "rank": 2, "domain": "capterra.com" }
  ],
  "brands_mentioned": ["HubSpot", "Pipedrive", "Zoho"],
  "country": "us",
  "latency_ms": 3180
}
Answer text Cited sources Brands mentioned Ranked domains Region targeting Batch prompts JSON & Markdown
What comes back

Everything ChatGPT put on screen — parsed into fields.

ChatGPT renders a wall of prose with linked citations and inline recommendations. cloro pulls it apart so each piece is its own typed value.

💬

The full answer

The complete response text, cleaned of UI chrome and ready to store, embed or diff. Ask for Markdown and you get headings and lists preserved for grounding a model.

clean prose Markdown option stable keys
🔗

Cited sources

Every link ChatGPT surfaced, in order, with domain and rank — so you can see which sites shape the answer.

🏷️

Brand mentions

The products and companies named in the reply, extracted as a list you can count and track over time.

🌍

Region-aware

Set a country and see the answer a user in that market would actually receive, not a generic one.

📦

Batch native

Push a whole prompt set in parallel and collect every structured record from a single job.

Why an API

Scraping ChatGPT by hand breaks constantly.

The app has no public read API, ships behind bot defenses and re-renders its markup on a whim. cloro carries that weight so your side stays a single request.

  • Rendering, rotation and retries handled upstream
  • Answers normalized into one schema across every engine
  • Citations and brands extracted, not left in raw HTML
  • Failed or empty pulls are never billed
See the schema
answerFull response text, chrome stripped
sources[]Ranked list of cited domains
brands_mentioned[]Products and companies named
countryMarket the prompt was run for
formatjson or markdown
latency_msTime to fetch and parse
The request lifecycle

One prompt in, a stored record out

01

Send the prompt

POST your prompt with an optional country and output format. One call handles a single prompt; a batch job handles thousands.

02

cloro asks ChatGPT

We run the prompt live in the right region, wait for the full answer to settle and capture the rendered result.

03

We parse it apart

The answer, its cited sources and every named brand are extracted and typed into the shared cloro schema.

04

You store & diff

Structured JSON or Markdown lands in your pipeline — ready to save, compare against last week and act on.

1endpoint for ChatGPT
3field groups: answer · sources · brands
2formats: JSON & Markdown
195+markets you can target
📊

Brand share of voice in ChatGPT

Run one prompt set weekly and count how often ChatGPT names each brand across every answer — a live read on who the model recommends.

Your brand
32%
Rival A
27%
Rival B
21%
Rival C
20%

example distribution — computed from your own prompt runs

What teams build

From raw answers to a visibility metric

  • GEO & SEO teams — prove whether your work lifts you inside ChatGPT answers
  • Brand & PR — watch how ChatGPT describes you and which rivals it names instead
  • Product & data — ground features on fresh, structured ChatGPT output
  • Agencies — report ChatGPT visibility across a client roster from one key
Track your prompts
ChatGPT API · FAQ

Good to know

What exactly do I get from a ChatGPT call?

The full answer text, an ordered list of the sources ChatGPT cited (domain and rank), and the brands or products it named — all in one structured record. Choose JSON for apps and databases, or Markdown when you're feeding a model.

Do I need a ChatGPT account or my own proxies?

No. cloro runs the prompt, handles rendering, rotation and retries, and returns parsed data. You authenticate with a single cloro API key — nothing else to set up.

Can I run many prompts at once?

Yes. Batch jobs push thousands of prompts in parallel across regions and return every structured record together, so you can cover a whole keyword or prompt set in one pass.

Can I target a specific country?

Yes. Set a country on the request and cloro captures the answer a user in that market would receive, so regional differences show up in your data.

How am I billed?

Per successful call. If a pull fails or comes back empty, it isn't charged, and your usage counter reflects exactly what you consumed.

Read ChatGPT like a database.

Get a key, POST your first prompt, and start turning ChatGPT answers into structured data you can track over time.

Get an API key Explore the docs