ChatGPT API · 100 free credits

Prompt in.
Answer out.

Q: Should I use this or AI extraction on the scrape API?

To pull fields out of a web page you are scraping, add ai_query or ai_schema to the scrape request: it extracts in the same call for 5 extra credits. Use this endpoint for everything else, such as text you already have, data from the TikTok, Instagram or Amazon APIs, or writing new copy.

One POST request with a prompt returns the answer as text. Summarize, classify, translate or reshape the data you collect, with the same API key and credit balance you already scrape with.

Get your free API key Read the docs

Requests served: 27M+
Credits per prompt: 10
Failed prompts: Free

curl -X POST https://scrapingbot.io/api/v1/chatgpt \
  -H "x-api-key: YOUR_API_KEY" \
  -H "content-type: application/json" \
  -d '{"prompt": "Summarize this review in one sentence: The battery easily lasts two days and the screen is sharp, but the camera struggles in low light and there is no charger in the box."}'

curl -X POST https://scrapingbot.io/api/v1/chatgpt \
  -H "x-api-key: YOUR_API_KEY" \
  -H "content-type: application/json" \
  -d '{"prompt": "Classify the sentiment of this comment as positive, negative or neutral. Reply with one word. Comment: Arrived two days early and works perfectly."}'

curl -X POST https://scrapingbot.io/api/v1/chatgpt \
  -H "x-api-key: YOUR_API_KEY" \
  -H "content-type: application/json" \
  -d '{"prompt": "Return only JSON with the keys name, price and rating. Text: Trail Runner 2 running shoes, now $129, rated 4.6 out of 5 by 2,300 buyers."}'

curl -X POST https://scrapingbot.io/api/v1/chatgpt \
  -H "x-api-key: YOUR_API_KEY" \
  -H "content-type: application/json" \
  -d '{"prompt": "Translate into Spanish and German: Free shipping on all orders over $50."}'

200 OK · 1.23 s · 10 credits application/json

{
  "success": true,
  "response": "Great battery life and a sharp screen, but weak low-light photos and no charger included.",
  "duration": "1.23",
  "statusCode": 200
}

{
  "success": true,
  "response": "positive",
  "duration": "0.64",
  "statusCode": 200
}

{
  "success": true,
  "response": "{\"name\": \"Trail Runner 2\", \"price\": 129, \"rating\": 4.6}",
  "duration": "0.92",
  "statusCode": 200
}

{
  "success": true,
  "response": "Spanish: Envío gratis en todos los pedidos superiores a 50 $.\nGerman: Kostenloser Versand für alle Bestellungen über 50 $.",
  "duration": "1.05",
  "statusCode": 200
}

The API

One endpoint.
One field.

Send a prompt, read the answer. There is nothing else to configure, and errors come back as JSON too, with success: false and a message you can log.

Authenticate with the x-api-key header, a Bearer token or an api_key query parameter, exactly as with every other ScrapingBot API.

Request

POST /api/v1/chatgpt

prompt required

stringYour instructions plus any text the model should work on.

Response

application/json

success

booleantrue when an answer came back.

response

stringThe answer, as plain text.

duration

stringSeconds the request took, such as "1.23".

statusCode

number200 on success.

Errors

success: false

{ "success": false, "error": "Missing required parameter: prompt" }

A missing prompt returns 400 and a low balance returns 402, both before anything is charged. If generation fails, the error comes back and the 10 credits are refunded automatically.

Works with your data

Scrape it, then make sense of it.

The same key and credit balance cover scraping and prompts, so a pipeline is two requests: collect the data, then ask about it.

Each prompt stands alone

There is no conversation memory. Put the instructions, the context and any earlier messages you need into the prompt.

Ask for the format you want

Need JSON, a single label or a bullet list? Say so in the prompt and parse response in your code.

Extracting from a web page?

Add ai_query to a website scrape instead. It extracts the fields in the same request for 5 extra credits.

Summarize a video’s comments · 11 credits

import requests

BASE = "https://scrapingbot.io/api/v1"
HEADERS = {"x-api-key": "YOUR_API_KEY"}

# 1. Pull comments from a TikTok video (1 credit)
video = "https://www.tiktok.com/@tiktok/video/7106594312292453675"
res = requests.post(f"{BASE}/tiktok", headers=HEADERS, json={
    "endpoint": "/comment/list",
    "params": {"url": video, "count": 50},
})
comments = [c["text"] for c in res.json()["data"]["comments"]]

# 2. Ask for the main themes (10 credits)
prompt = "List the three main themes in these comments:\n"
prompt += "\n".join(comments)
res = requests.post(f"{BASE}/chatgpt", headers=HEADERS,
                    json={"prompt": prompt})
print(res.json()["response"])

Use cases

What teams use it for.

Most prompts fall into one of these four jobs.

Summarize

Condense reviews, comments, articles or support threads into a line or a paragraph.

Classify and tag

Label sentiment, topic, language or intent so your data is easy to filter and count.

Extract to JSON

Ask for specific fields as JSON and parse the answer straight into your database.

Translate and rewrite

Localize product copy, or rewrite text in a different tone, length or reading level.

Billing

You pay for answers,
not attempts.

A ChatGPT prompt costs 10 credits, and only when it succeeds. Short or long, every prompt costs the same.

Failed prompts are refundedErrors and timeouts are credited back automatically, and requests with missing or invalid parameters are never charged.
100 free credits to startThat covers 10 ChatGPT prompts on the free plan. No credit card needed to sign up.
See the cost of every promptYour dashboard logs every request with its status, duration and the credits it used.
One balance for every APIThe same credits work for website scraping, TikTok, Instagram, Google, Amazon and ChatGPT. Paid plans refresh them every billing cycle.

See plans and credits

Every successful ChatGPT prompt

10credits

Free to start: 10 prompts; from 100 free credits
Failed prompts: 0 credits; refunded automatically

Questions, answered

How do I call it?

Send a POST request to https://scrapingbot.io/api/v1/chatgpt with your API key in the x-api-key header and a JSON body like {"prompt": "…"}. The answer comes back as text in the response field.

Does it remember earlier messages?

No. Every request is independent. Include any instructions, context or earlier conversation you need in the prompt itself.

Can I choose the model or set options like temperature?

No. The endpoint takes a prompt and nothing else, so there is nothing to tune. Ask for the tone, length or format you want in the prompt.

Can I get JSON back?

Yes. Ask for it in the prompt, for example "Return only JSON with the keys name and price". The answer arrives as a string in response, so parse it in your code.

Should I use this or AI extraction on the scrape API?

To pull fields out of a web page you are scraping, add ai_query or ai_schema to the scrape request: it extracts in the same call for 5 extra credits. Use this endpoint for everything else, such as text you already have, data from the TikTok, Instagram or Amazon APIs, or writing new copy.

What does it cost?

10 credits per successful prompt, however long it is. Errors are refunded automatically, and a request without a prompt is rejected without charge. New accounts get 100 free credits, enough for 10 prompts.

Where can I try it before writing code?

Every account has a ChatGPT playground in the dashboard where you can test prompts and copy the request. The docs include examples for common tasks.

Prompt in.Answer out.

One endpoint. One field.