# Firecrawl Search, collect, and structure web content for AI agents and application data pipelines. - Findings24: - ウェブサイト: - 料金: Freemium - カテゴリー: AIデータ抽出, AI検索API, AIウェブスクレイパー ## Firecrawlとは？ Firecrawl gives developers APIs for finding web pages and collecting content as Markdown, HTML, or structured JSON. Applications can scrape a page, crawl linked pages, or map URLs before selecting what to fetch. The hosted service also provides AI-directed browser interaction for tasks that require navigation. An open-source core is available for self-hosting, with a different operating and feature scope from Firecrawl Cloud. ## Firecrawlでできること ### Search and collection Find search results, optionally retrieve their content, scrape individual pages, or crawl a site. ### URL discovery Map a site's URLs before choosing pages to collect. Mapping discovers links rather than downloading each page's content. ### Structured extraction Return cleaned Markdown or HTML, or use LLM-backed formats to extract JSON fields and relevant passages. ### Dynamic pages and browser interaction Render JavaScript content and use hosted browser sessions for clicking, typing, scrolling, and navigating with code or AI prompts. ### Developer access Connect through the HTTP API, Python or Node.js SDKs, CLI, and MCP-compatible agents. Core code is primarily AGPL-3.0; SDKs and some UI components use MIT licenses. ## 利用シーン ### Website knowledge retrieval Collect documentation and reference pages for search indexes or retrieval-based assistants. ### Research workflows Discover pages, retrieve supporting content, and extract fields for comparison or analysis. ### Product and company data Gather published product details and company information for application-managed catalogs or records. ## よくある質問 ### What do the free and entry paid plans include? Free includes 1,000 credits per month and two concurrent browsers, without a card. Hobby includes 5,000 credits monthly and five concurrent browsers for $19 USD per month or $190 per year. API request-rate limits are separate from browser concurrency. ### How do operations consume credits? Basic Scrape and Crawl cost one credit per page; Map costs one per call. Search costs two per ten results, with extra costs for scraping those results. JSON extraction adds four credits per page. Interact costs two credits per minute for code-only sessions or seven with an AI prompt, with a one-minute minimum; additional options can increase cost. ### Can paid usage continue after credits run out? Paid plans can add credits automatically in $5 increments. Set a monthly pay-as-you-go cap, or set it to zero to disable automatic purchases. All endpoints use the same credit balance. ### Does self-hosting include every hosted feature? No. The core scraping APIs are available, but LLM-backed formats require your own provider configuration. Agent, Browser, Interact, managed dashboards, and enterprise controls are outside the default open-source stack. You operate the infrastructure and pay its provider costs. ### Where is zero data retention available? Enterprise offers zero retention of scraped content for approved use cases. It is not a general promise that every plan or all account and billing information is never stored. ## クイック情報 - **Deployment**: Hosted API or self-hosted open-source core - **Outputs**: Markdown, HTML, JSON, and other requested formats - **Hosted billing**: Shared credits with operation-specific costs - **Concurrency unit**: Browsers processing pages simultaneously