# Firecrawl > Web data API that searches, scrapes, crawls and interacts with websites and returns clean output for AI applications. Firecrawl is a web data API that searches the web, scrapes pages, crawls sites and interacts with pages on behalf of AI applications. Firecrawl is used by sending a query or URL through its API, SDK, CLI or MCP server and receiving Markdown, HTML, screenshots or structured JSON. Firecrawl is open source and includes 1,000 free credits monthly. **Best for:** Developers building AI agents, research tools or retrieval pipelines that need current web content in a model-friendly format. - Official site: https://www.firecrawl.dev - Listed in: [Web Scraping & Data Extraction](https://trendhub.net/ai-tools/category/web-scraping), [AI Agents & Automation](https://trendhub.net/ai-tools/category/ai-agents), [Developer API](https://trendhub.net/ai-tools/category/developer-api) - Last checked: 2026-10-09 (facts taken from firecrawl.dev) - Source page: https://trendhub.net/ai-tools/firecrawl-web-scraping-search-api-ai ## Key facts - Platforms: API, MCP, Web - Pricing model: Freemium - Starting price: $16 per month (Hobby plan billed annually; $19 per month when billed monthly) - Free plan: Yes - API: Yes - Made by: Firecrawl ## What is Firecrawl? Firecrawl is a web data API that lets software, and AI agents in particular, find and read content on the live web. It has three main functions. Search takes a query and returns results with the full page text attached. Scrape takes a URL and returns the page as Markdown, HTML, a screenshot, metadata or structured JSON that follows a schema you supply. Interact drives a page by clicking, typing and navigating when data sits behind forms or pagination. Crawl and Map endpoints cover whole sites, and JavaScript rendering is handled automatically. Firecrawl is open source, with a hosted version that adds the company's own proxy and rendering infrastructure and a dashboard. It ships SDKs for several languages, a CLI and an MCP server for tools such as Cursor and Claude. Billing is credit based, with a free monthly allowance. ## Key features - **Search**: Runs a web query and returns matching results together with each page's content in Markdown, so no second scraping call is needed. - **Scrape**: Fetches a single URL and outputs Markdown, raw HTML, a screenshot, metadata or JSON, leaving out navigation, footers and adverts. - **Crawl and Map**: Follows links from a starting address across a site, with controls for depth, page limits and path filters, and respects robots.txt rules for its agent. - **Interact**: Operates a loaded page through AI prompts or code, performing clicks, form entry, scrolling and key presses. - **Structured Extraction**: Accepts a JSON schema with a scrape request and returns data in that shape, for example product listings or contact details. - **Document Parsing**: Reads content from files such as PDF and DOCX in addition to web pages. ## Use cases - **Deep Research Agents**: Gather papers, news and reports on a topic for an agent that writes research summaries. - **Retrieval for AI Chat**: Ingest website content into a knowledge base so an assistant can answer with current information. - **Lead Enrichment**: Add company and contact details from the web to existing sales records. - **User Onboarding**: Read a new customer's website to pre-fill their account or first project. - **Web Tools for Coding Agents**: Give Cursor, Claude Code or Windsurf search and scrape abilities through the CLI or MCP server. ## Pricing | Plan | Price | Billing | Includes | | --- | --- | --- | --- | | Free | Free | per month, no card required | 1,000 credits per month; About 1,000 scraped pages or 500 searches; 2 concurrent requests; Low rate limits | | Hobby | $16 | per month, billed annually; $19 per month when billed monthly | 5,000 credits per month; About 5,000 scraped pages or 2,500 searches; 5 concurrent requests; Basic support | | Standard | $83 | per month, billed annually | 100,000 credits per month; About 100,000 scraped pages or 50,000 searches; Standard support | | Enterprise | Custom | contact sales, custom | Custom; Dedicated support and SLA; Zero-data retention; SSO and advanced security | Prices as stated in the text of the official pricing page, which says prices are in USD and shown per month with annual billing. Growth and Scale plans also exist; their amounts are not listed here because the page text does not pair them with the plan names unambiguously. Paid plans can add credits in 5 USD increments. Checked 2026-10-09 at https://www.firecrawl.dev/pricing ## Things to know - **Credits differ by endpoint**: A basic scrape, crawl or map uses one credit per page, search uses two credits per ten results, and JSON-style formats add four credits per page. - **Rollover only on top tiers**: Unused credits carry over on Scale and Enterprise; credits on Hobby, Standard and Growth do not. - **Automatic top-ups require a paid plan**: When a paid plan's credits run out, more are added in 5 USD steps up to a monthly cap you choose; the option can be switched off. - **Error pages can still cost a credit**: A scrape with no result is not charged, but a page that answers with a status such as 403 or 404 is returned and billed one credit. ## FAQs ### What is Firecrawl? Firecrawl is a web data API that searches, scrapes, crawls and interacts with websites and returns the content in formats suited to language models. ### What is Firecrawl used for? Firecrawl is used to supply AI applications with web content, for example in research agents, retrieval pipelines, lead enrichment, price monitoring and competitive analysis. ### Is Firecrawl free? Yes, Firecrawl has a free tier with 1,000 credits each month, roughly 1,000 scraped pages, and it does not ask for a card. Larger volumes need a paid plan. ### How much does Firecrawl cost? Firecrawl's Hobby plan costs $19 per month billed monthly or $16 per month billed annually for 5,000 credits, and Standard is $83 per month billed annually for 100,000 credits. Prices are stated in USD; Growth, Scale and Enterprise tiers sit above these. ### Is Firecrawl open source? Yes. Firecrawl publishes its code on GitHub. The hosted service adds proprietary infrastructure for proxies and rendering, page interaction and an analytics dashboard. ### Does Firecrawl work with AI agents and MCP? Yes. Firecrawl has an official MCP server for clients such as Cursor, Claude and Windsurf, a command-line tool, and ready-made skills that set it up inside Claude Code and Codex. ### Which programming languages does Firecrawl support? Firecrawl provides SDKs in six languages (Python, Go, Rust, Java, Elixir and Node.js), and its REST API can be called from any other language. ### What output formats does Firecrawl return? Firecrawl returns Markdown by default and can also provide screenshots, raw HTML, metadata about the page and JSON structured by a schema. --- Source: https://trendhub.net/ai-tools/firecrawl-web-scraping-search-api-ai (trendhub)