Skip to content

Capture Public Grok Shared-Chat Pages Through One API Call

Public Grok shared chats are a useful record of how people prompt a model, which answers get shared, and which sources a response cites.

OUTPUT FORMATSHTMLJSONMarkdown

Collecting those pages by hand does not scale, and a plain HTTP fetch often returns an empty shell because the conversation is rendered client-side. This page explains what an API-based renderer returns for Grok shared-chat URLs, which request options map to which extraction goal, and how the credit model turns into a monthly request budget.

What a public Grok shared chat exposes

A publicly shared Grok conversation is a rendered page, not a database row. Once the page's JavaScript has run, four kinds of content are typically available to a parser: the prompt text the user submitted, the response text Grok produced, the citation URLs and references attached to the answer, and the turn-by-turn order that links each prompt to its reply. That structure is what makes shared chats useful for prompt-pattern research, brand-mention monitoring, and citation auditing. Anything behind a login is out of scope; the target must be a page a logged-out visitor can open.

Three workflows that justify the call

The recurring uses cluster into three jobs. First, prompt and model research: collect many shared conversations to see how users phrase requests and which model variants appear. Second, brand and competitor monitoring: scan shared chats for your product name or a rival's and record where it surfaces in AI conversation. Third, citation auditing: pull the source URLs Grok cites so you can see which domains the answers lean on and check factual grounding. Each job needs the same raw material but a different downstream schema, which is why selector-based and AI-based extraction both matter.

Matching request options to extraction goals

The API accepts a Grok URL and a set of options. The table below is a reader guide, not a new contract; the exact field names and credit rules live in the protected controls module delivered with this page.

Turning credits into a request budget

Credit cost per request depends on which options you enable, so the same monthly allowance buys very different volumes. Requests start at 1 credit each. Enabling JavaScript rendering costs 5 credits per successful request, so a JS-rendered call totals 6 credits. Premium proxies cost 10 credits with JS rendering disabled or 25 with it enabled, and stealth_proxy costs 75 credits. On a 75,000-credit allowance, an all-JS-rendered workload at 6 credits per call is roughly 12,500 requests, an all-premium-proxy-plus-JS workload at 25 credits is about 3,000 requests, and an all-stealth workload at 75 credits is about 1,000 requests. In practice you would reserve stealth for the pages that actually block and run the bulk on plain JS rendering. Treat these as arithmetic on the stated per-request costs, not a promise about how many Grok pages will succeed at each tier.

Three different failures, three different fixes

When a Grok scrape comes back wrong, the cause is usually one of three things and the fix differs each time. An empty or blocked response means the render did not get through the anti-bot layer: raise the proxy tier or enable stealth rather than rewriting selectors. A full page with missing fields means the selector or extraction rule no longer matches the markup: inspect the rendered HTML and update extract_rules, or switch to ai_extract_rules for a natural-language description of the fields. A page that only exists after login is not a scraping failure at all; it is out of scope and should be excluded from the crawl list.

From signup to first parsed response

The path is short: create an account to get an API key and starting credits, install an SDK for your stack or call the endpoint directly, then send a GET request with the Grok URL. The renderer executes the page's JavaScript and rotates proxies, returning rendered HTML you can parse. SDKs are published for Python, Node.js, Go, PHP, Ruby and Java, and plain cURL works as well. A dashboard request builder lets you test parameters without writing code first.

Plan tiers as a dated reference

The plan table below records the published tiers and their credit and concurrency allowances as captured. Prices are exclusive of VAT. Treat the figures as a historical reference point and confirm current pricing and terms before committing to a plan; the per-request credit costs in the controls module are the part that determines your real monthly spend.

Request option to extraction goal
OptionWhat it doesWhen to reach for it
render_jsExecutes the page's JavaScript before returning contentDefault for Grok shared chats, since the conversation is client-rendered
stealth_proxyRoutes through a stealth proxy tier for harder anti-bot pagesWhen a normal render returns a block or empty shell
country_codeSets the geolocation of the requestWhen content or availability varies by region
extract_rulesCSS or XPath selectors that pull structured fieldsWhen the markup is stable and you want deterministic JSON
ai_extract_rulesNatural-language description of fields to extractWhen markup shifts often or you want to describe fields in plain language

Option names are exact technical identifiers; behavior descriptions are a reader guide and the protected controls module is authoritative for credit rules.

Published plan tiers (captured reference)
PlanPrice / monthAPI creditsConcurrent requests
Hobby$1975,00025
Freelance$49250,00050
Startup$991,000,000100
Business$2493,000,000200
Business +$5998,000,000400
Enterprise 14$99914,000,000500
Enterprise 24$1,59924,000,000600
Enterprise 41$2,39941,000,000700
Enterprise 69$3,59969,000,000800
Enterprise 120$5,799120,000,000900

Prices exclusive of VAT. Captured reference only; confirm current pricing and terms before purchase. All tiers include JavaScript rendering, rotating and premium proxies, geotargeting, and screenshots/extraction rules; priority email support, dedicated account manager, and team management are included from Business upward.

Frequently asked questions

Do I need a Grok account to scrape public shared chats?

No external account is needed. You sign up for an API key, which comes with 1,000 free credits and no card requirement, then send the shared-chat URL; the renderer and proxy handling are on the API side.

Is scraping Grok shared chats legal?

Public-facing pages are generally the ones you can collect, but you should read the target site's terms of service before deploying and keep post-login content out of scope. This is a general caution, not legal advice for your jurisdiction.

What can I extract from a Grok shared chat?

Anything that renders on the public page. Use extract_rules with CSS selectors for deterministic fields, or ai_extract_rules to describe fields in natural language and get structured JSON back.

How do I make my first Grok request?

Sign up, copy your API key, and send a GET request with the Grok URL. The service renders the page, rotates proxies, and returns the content; 1,000 free credits are available for testing.

Which languages can I use?

SDKs are published for Python, Node.js, Go, PHP, Ruby and Java, and plain cURL works too, so you can stay in whatever stack you already run.

How are credits counted per Grok request?

Requests start at 1 credit each. JavaScript rendering adds 5 credits per successful request, so a JS-rendered call totals 6 credits. Premium proxies cost 10 credits with JS rendering disabled or 25 with it enabled, and stealth_proxy costs 75 credits. Credits are valid for one month and do not roll over, so size your plan against the option mix you actually use.