Site integrations
ChatGPT
ChatGPT answers: the answer a signed-out visitor gets for a prompt with web search on, and the citations behind it.
Call these actions with your String API key; see Authentication. Every call answers with the envelope described in Responses & errors, whose data holds the rows below.
- Site:
chatgpt - Version:
1.0.0 - Reads from:
chatgpt.com
| Action | Description | Listed price |
|---|---|---|
answer | Ask ChatGPT one question as a signed-out visitor with web search on and get the answer it writes: markdown text with inline citation markers, the sources it cites, whether a web search ran and how many sites it searched, the conversation id, and the persona and web build the answer was served under. | Billed per request at your plan's rates |
answer
Ask ChatGPT one question as a signed-out visitor with web search on and get the answer it writes: markdown text with inline citation markers, the sources it cites, whether a web search ran and how many sites it searched, the conversation id, and the persona and web build the answer was served under.
POST https://request.usestring.ai/v1/integrations/chatgpt/answerInput
Unknown fields are rejected.
| Field | Type | Description |
|---|---|---|
query | string | Required. The question to ask ChatGPT. At least 1 character. Matches \S. |
Output
data is an array of rows, one per result. Each row has:
| Field | Type | Description |
|---|---|---|
query | string | — |
answerMarkdown | string | The answer's markdown text, with the inline citation markers kept verbatim. |
conversationId | string | null | The conversation the answer belongs to. |
webSearchRan | boolean | Whether a web search ran for the answer. |
searchedSiteCount | integer | null | How many sites the answer reports searching; null when the stream does not say. |
persona | string | The client persona the answer was served under, such as chatgpt-noauth. |
clientBuild | string | The chatgpt.com web build that produced the answer. |
citations | object[] | The sources the answer shows, in the order it shows them: the ones its inline citation markers point at (Cite) and the links on the cards it shows (Link) |
citations[].position | integer | Position in the answer's source list, from 1. |
citations[].kind | string | Cite for a source an inline citation marker points at, Link for a link on a card the answer shows, such as a product's website. |
citations[].refId | string | null | The id an inline citation marker in answerMarkdown names this source by, such as turn0search3; null for a Link. |
citations[].url | string | — |
citations[].title | string | null | — |
citations[].sourceLabel | string | null | The site name shown for the source. |
citations[].snippet | string | null | The passage the source contributed, when the stream carries one. |
citations[].domain | string | The source URL's host, without a leading www. |
citations[].publishedTimestamp | integer | null | The source's publication time in unix seconds, when the stream carries it. |
Example
curl https://request.usestring.ai/v1/integrations/chatgpt/answer \
-H "Authorization: Bearer $STRING_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "What are the best web data extraction providers?"
}'data starts with these rows:
[
{
"query": "What are the best web data extraction providers?",
"answerMarkdown": "The best web data extraction providers depend on whether you need structured data from specific websites, large-scale scraping, full-site crawling for AI, or a managed service that delivers datasets without your team maintaining scrapers.\n\nFor most teams, I'd start by evaluating **Apify, Bright Data, Zyte, Firecrawl, and ScrapingBee**\\. They cover different needs, from ready-made scrapers to enterprise-grade extraction and AI-ready content. citeturn920565search12turn920565search13turn920565search15\n\n## Best web data extraction providers\n\nShortlist by use case · October 2026\n\nimage_group{\"query\":[\"Apify web scraping platform logo\"],\"aspect_ratio\":\"1:1\",\"layout\":\"carousel\",\"num_per_query\":1}\n\n## 1\\. entity[\"company\",\"Apify\"]\n\n(Best overall for flexible workflows)\n\nA developer-friendly platform with a marketplace of ready-made scrapers, customizable extraction jobs, scheduling, a…",
"conversationId": "6ad7bc8d-b10c-83ea-844e-5fd01073389a",
"webSearchRan": true,
"searchedSiteCount": 16,
"persona": "chatgpt-noauth",
"clientBuild": "prod-acfed2b7fc7a93893746c1bcab9c52f5b379caa6",
"citations": [
{
"position": 1,
"kind": "Cite",
"refId": "turn920565search12",
"url": "https://brightdata.com/blog/web-data/best-web-scraping-apis?utm_source=chatgpt.com",
"title": "The 9 Best Web Scraping APIs & Tools in 2026",
"sourceLabel": "Bright Data",
"snippet": null,
"domain": "brightdata.com",
"publishedTimestamp": null
},
{
"position": 2,
"kind": "Cite",
"refId": "turn920565search13",
"url": "https://crawl.sarv.com/blog/best-web-scraping-apis?utm_source=chatgpt.com",
"title": "Best Web Scraping APIs in 2026: 8 Compared",
"sourceLabel": "SarvCrawl",
"snippet": null,
"domain": "crawl.sarv.com",
"publishedTimestamp": 1789430400
},
{
"position": 3,
"kind": "Cite",
"refId": "turn920565search15",
"url": "https://www.itechguides.com/11-best-web-scraping-services-in-the-usa-in-2026/?utm_source=chatgpt.com",
"title": "11 Best Web Scraping Services in the USA (2026)",
"sourceLabel": "iTechGuides",
"snippet": null,
"domain": "itechguides.com",
"publishedTimestamp": 1790726400
}
]
}
]Requests and billing
- Requests: 5 Web Access requests per call.
- Billing:
passthrough. Each request the action makes bills at its own rate on your plan; there is no separate charge for the action. See Pricing.