Firecrawl is a web-scraping & crawling API built for LLMs: give it a URL and it returns clean markdown (handling JavaScript, proxies and anti-bot for you), or crawl a whole site for RAG. Here's the credit-based pricing, the free allowance, and how to start.
| Plan | Price | Credits (≈ pages) |
|---|---|---|
| Free free tier | $0 | ~500 one-time credits to test |
| Hobby | from ~$16/mo | ~3,000 credits / mo |
| Standard | from ~$83/mo | ~100,000 credits / mo, higher concurrency |
| Growth / Enterprise | custom | high volume, more concurrency, support |
Firecrawl gives a one-time free credit allowance (around 500 credits) — enough to scrape a few hundred pages and prove out a RAG ingestion pipeline before paying. It's a trial pool, not a monthly free quota, so plan to upgrade once you're running real crawls. For small recurring jobs, a cheaper raw scraper plus your own markdown conversion can undercut it.
1. Sign up at firecrawl.dev.
2. Open the dashboard — you'll see your free credits.
3. Go to API Keys and copy your key (starts with fc-).
4. Call the scrape endpoint with a Bearer token.
Scrape one page to markdown:
If you only need raw HTML or search results, a generic scraper is cheaper: ScraperAPI and Bright Data handle proxies at lower per-request cost, and SerpAPI returns search results as JSON. Firecrawl's edge is the LLM-ready markdown and whole-site crawl — worth it when you're feeding a RAG pipeline and don't want to write HTML cleaning. Pair it with an embeddings store like Pinecone and an LLM like OpenAI or Claude.
Yes — a one-time ~500-credit allowance to test scraping and crawling. After that you move to a monthly plan. It's a trial pool, not a recurring free quota.
Sign up at firecrawl.dev, open the dashboard, go to API Keys, and copy the key (it starts with fc-). Use it as a Bearer token on the /scrape or /crawl endpoint.
Roughly one credit per page scraped or crawled. Extra features like JavaScript rendering, screenshots or structured extraction can cost more credits per page, so a crawl of 1,000 pages with extraction uses well over 1,000 credits.
Yes — that's its main use case. The crawl endpoint walks a site and returns clean markdown per page, which you can chunk and embed into a vector database for retrieval-augmented generation.
Not affiliated with Firecrawl. Prices are reference estimates — always verify on the official pricing page.