Skip to content

GrubHub Scraper API: Menu, Price and Rating Data in One Request

GrubHub pages combine JavaScript rendering, rotating proxies and bot detection, so a plain HTTP request often returns a challenge page instead of a menu.

OUTPUT FORMATSHTMLJSONMarkdown

This page explains how a managed scraping API handles those layers for you, what each request option does, how credits are consumed, and where the practical limits sit — so you can decide whether to build on it or keep your own stack.

What the API handles for you

You send a GrubHub restaurant or menu URL. The service renders the page's JavaScript, rotates proxies, and attempts to bypass bot detection before returning either the rendered HTML or structured JSON. The practical result is that menu items with prices and modifiers, restaurant ratings, review counts, delivery-time estimates, delivery fees and minimum orders can be collected without you operating headless browsers or a proxy pool.

Three research questions tend to drive GrubHub collection, and each maps to a different set of fields to capture:

  • What is the delivery markup, and where does it vary? The analytical question is the gap between a chain's delivery-menu price and its dine-in price, sliced by zip. Capture item name, price and modifier surcharges per restaurant, and record the zip and capture time alongside each price — without those two fields the comparison is not defensible, because delivery menus change.
  • What does a full menu look like as a record? Capture category, item, price and modifier groups so the menu can be joined to other datasets. Decide up front whether you need modifiers or just base items — modifiers multiply row counts.
  • How do ratings and review counts compare across competitors? Capture rating and review count per restaurant, and note that a rating without its review count is easy to misread.

If your project only needs a handful of pages once, a manual approach may be enough. The API becomes worthwhile when you need repeatable collection across many zips or restaurants and do not want to maintain the anti-bot layer yourself.

Beyond the core scrape, the same service exposes adjacent capabilities that often sit next to a GrubHub project:

  • AI data extraction — describe the restaurant fields you want (menus, prices, reviews) and let AI extraction return them.
  • Page screenshots — capture accurate full-page images of GrubHub pages as a visual record for scraper projects.
  • Google search-result extraction — pull GrubHub results straight from the SERP to gather competitive data.
  • JavaScript scenarios — automate clicks, scrolls and typing so all necessary content is fully loaded before capture.
  • Structured Markdown — receive clean, structured Markdown for menu items, reviews and pricing, ready for LLM pipelines.
  • n8n no-code automation — build automated workflows that call the API without writing code.

How a request flows

The workflow has four stages:

  1. Account and key. Create an account to receive an API key and starting credits.
  2. Client. Install an SDK for Python, Node.js, Go, PHP, Ruby or Java, or call the endpoint with cURL. Each scrape becomes a single function call.
  3. Request. Pass a GrubHub URL and your API key. The service renders the page and attempts the bot bypass.
  4. Response. Receive parsed HTML or structured JSON.

When the default request is not enough, advanced parameters take over:

  • render_js — execute page JavaScript.
  • stealth_proxy — route through stealth proxies for harder bot protection.
  • country_code — geotarget the request.
  • extract_rules — map CSS selectors to field names for structured output.
  • ai_extract_rules — describe the fields you want in natural language and let AI extraction return them.

A useful way to think about the controls: render_js and stealth_proxy change how the page is fetched; extract_rules and ai_extract_rules change how the response is shaped. Choosing the shaping option first usually saves credits, because you avoid fetching more than you need.

Credits, concurrency and rate limits

Credit cost depends on the options you enable, not on the target site. The documented per-request costs are:

  • Base requests start at 1 credit each.
  • JavaScript rendering costs 5 credits per successful request.
  • Premium proxies cost 10 credits without JS rendering, or 25 credits with JS rendering. The tooltip presents the 25 figure as the premium-proxy-with-JS price; the FAQ separately lists 5 credits for JS rendering and 25 for premium_proxy with JS. Treat the 25 figure as inclusive per the tooltip, but note the source ambiguity rather than assuming it is settled.
  • Stealth proxy costs 75 credits.

API credits are valid for one month and do not roll over. Concurrency is limited per plan — the number of requests that can be sent at the same time — rather than by a per-target rate limit. If GrubHub throttles aggressive patterns, space out requests or use multiple sessions.

Failure distinction. Three different problems can look similar in logs:

  • *Concurrency limit reached* — your own plan's simultaneous-request ceiling. Queue or upgrade.
  • *Target throttling* — GrubHub returning challenges or slow responses because of request patterns. Slow down or vary sessions.
  • *Credit exhaustion* — the account has run out of monthly credits. Check the dashboard before assuming a scraping failure.

A worked example: suppose you want to compare menu prices for 50 restaurants across 5 zip codes, and each restaurant page needs JS rendering plus a premium proxy. At 25 credits per request (the tooltip's inclusive reading), 250 requests consume 6,250 credits. That fits comfortably inside the entry-level plan's 75,000 monthly credits, leaving room for retries and additional pages. If the FAQ's additive reading applied instead, the per-request cost would be 30 credits and the total 7,500 — still within the entry tier. If you instead used stealth proxy at 75 credits per request, the same 250 requests would consume 18,750 credits — a materially different budget.

What comes back

The response is either rendered HTML or structured JSON, depending on your request. With extract_rules, you supply CSS selectors mapped to field names and receive parsed JSON keyed by those names. With ai_extract_rules, you describe the fields in natural language and the AI extraction layer returns them.

For a GrubHub restaurant page, the fields the service reports as available include menu items with prices and modifiers, restaurant rating, review count, delivery-time estimate, delivery fee and minimum order. Which of those appear in a given response depends on the page and the extraction rules you set — treat the list as the target schema, not a guarantee for every URL.

If you need the page as clean Markdown instead of JSON, that is a separate output mode. Screenshots are also available when you need a visual record of the rendered page.

Decision checklist before you build

Use these questions to decide whether the managed API fits your project:

  • Volume. How many pages per month, and how many credits per request will your options consume? Multiply before choosing a plan.
  • Freshness. Do you need current menu and price data, or is a periodic snapshot enough? Credit validity is monthly, so unused credits do not carry over.
  • Legal review. Public-facing pages can usually be scraped, but review GrubHub's terms of service before deploying and avoid post-login content.
  • Output format. Do you need JSON for a pipeline, Markdown for an LLM, or screenshots for audit? Each has a different request shape.
  • Failure handling. Have you planned for concurrency limits, target throttling and credit exhaustion as separate conditions?
  • Stack fit. Does an SDK exist for your language, or will you call the endpoint directly with cURL?

If most answers point to repeatable, structured collection at moderate volume, the managed route removes the proxy and browser maintenance. If you need only occasional pages or already operate a reliable anti-bot stack, a self-built approach may be simpler.

Plan reference

The plan structure below reflects the captured pricing page at the time of collection. Prices and credit allowances can change, so confirm current terms on the pricing page before committing. All prices are exclusive of VAT.

Standard plans range from a Hobby tier at $19/month with 75,000 credits and 25 concurrent requests, up to a Business + tier at $599/month with 8,000,000 credits and 400 concurrent requests. Enterprise plans range from Enterprise 14 at $999/month with 14,000,000 credits and 500 concurrent requests, up to Enterprise 120 at $5,799/month with 120,000,000 credits and 900 concurrent requests. JavaScript rendering, rotating and premium proxies, geotargeting, and screenshots and extraction rules are included across the listed tiers. Priority email support, a dedicated account manager and team management appear only from the Business tier upward.

A free starting allowance is available without a credit card, which is the lowest-risk way to test whether the request options and output shape fit your pipeline.

Frequently asked questions

Do I need a GrubHub account to use the scraper?

No. You need an account with the scraping service to get an API key, but you do not need to log in to GrubHub. Send the public URL and the service handles the rest. Avoid post-login content in any case.

Is scraping GrubHub legal?

Public-facing pages can usually be scraped, but the legal position depends on jurisdiction and on GrubHub's terms of service. Review those terms before deploying, and do not collect post-login content. This page is not legal advice.

Is there a rate limit per target site?

The API applies concurrency limits per plan rather than a per-target rate limit. If GrubHub throttles aggressive request patterns, space out requests or use multiple sessions. Concurrency is the number of requests that can be in flight at the same time.

How do I get structured JSON from a GrubHub page?

Use extract_rules with CSS selectors mapped to field names, or ai_extract_rules with a natural-language description of the fields you want. The response comes back as parsed JSON.

Which programming languages are supported?

SDKs are published for Python, Node.js, Go, PHP, Ruby and Java. Plain cURL also works, so you can integrate from any stack that can make an HTTP request.

How much does a GrubHub scrape cost in credits?

Base requests start at 1 credit. JavaScript rendering costs 5 credits per successful request. Premium proxies cost 10 credits without JS rendering or 25 with it; the 25 figure is presented as the premium-proxy-with-JS price, while the JS rendering charge is listed separately at 5 credits, so treat the inclusive reading as the documented tooltip rather than a settled rule. Stealth proxy costs 75 credits. Paid plans start at $19/month, and a free starting allowance is available without a card. Credits are valid for one month and do not roll over.