Crw
Web scraping and search API for AI agents — scrape, crawl, map, extract, and monitor the web from one REST endpoint.
Very few scraping vendors publish a dataset, a reproduction command, and their losses alongside their wins — fastCRW does, and the flat 1 credit per page on every renderer makes cost forecasting trivial. Pick it when you want an auditable, self-hostable crawl layer with a real MCP install path; pick Firecrawl or Tavily instead if you need a GUI-first workflow or a big-name compliance packet. Budget on the yearly Hobby/Standard figures, not the monthly headline.
Verified 4d ago · liveness 76/100 · cite: rightaichoice.com/tools/crw
- Agent developers who want search, scrape, crawl, extract, and monitor behind one API key and an MCP server
- RAG teams that need clean markdown at a flat per-page credit instead of token-metered scraping
- Teams that must self-host the crawl layer for data residency or audit reasons
- Engineers migrating off Firecrawl, Tavily, Exa, or Parallel.ai who want a reproducible benchmark before switching
- Non-technical teams wanting a point-and-click visual scraper with no API or CLI
- Buyers whose procurement gate is a SOC 2 or HIPAA report rather than an AGPL-3.0 license file and public benchmarks
- High-volume structured extraction pipelines that haven't modeled Extract's token-based metering on paid plans
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip fastCRW if you need a point-and-click visual scraper with no API, CLI, or MCP config, or if your procurement process requires a SOC 2 or HIPAA report in hand before you can start.
Extract is metered by token usage on top of your page credits, so a pipeline that scrapes at $0.00069 per page on Standard can still run a second, usage-based bill once you turn on schema extraction.
Published plans run Free, then Hobby at $11/mo billed $132/yr, Standard at $59/mo billed $708/yr, Growth at $237/mo billed $2,844/yr, and Scale at $467/mo billed $5,604/yr, with Enterprise custom. At roughly $0.00069 per page on Standard you are priced below Firecrawl's managed tiers for comparable scrape volume, and Standard's 100,000 credits a month covers most RAG builds; solo developers should start on Hobby, while teams past a million pages a month should price Scale against self-hosting
In short
Crw — Web scraping and search API for AI agents — scrape, crawl, map, extract, and monitor the web from one REST endpoint. Best for Agent developers who want search, scrape, crawl, extract, and monitor behind one API key and an MCP server, RAG teams that need clean markdown at a flat per-page credit instead of token-metered scraping, Teams that must self-host the crawl layer for data residency or audit reasons. Free to start; paid plans from $11/mo.
What's new in Crw
Checked 2 days agoAcross the latest 4 updates: 1 changelog entry and 3 news mentions.
fastCRW claims 90% on rebuilt AA-Omniscience search benchmark
fastCRW reports answering 90.0% of 600 AA-Omniscience questions correctly after rebuilding Artificial Analysis's harness and reproducing another provider's score within 1.5 points.
fastCRW pitches itself as a Parallel Extract API alternative
A guide for converting URLs to LLM-ready markdown that cites open-core code, self-hosting, a REST API, CLI, and MCP server as the differentiators against Parallel's Extract API.
fastCRW publishes Parallel.ai comparison covering search, extract, monitor
A surface-by-surface comparison of Parallel.ai's Search and Task APIs against fastCRW's search, extract, and monitor endpoints, with benchmark numbers and listed gaps.
Capacity rollout: sustained concurrency may run below published plan targets
fastCRW notes that during the capacity rollout sustained concurrent throughput may be lower than each plan's provisioned figure, and requests over live capacity receive a 503 with a Retry-After header.
What people actually say about Crw — is it worth it?
We scanned public community sources for Crw on Jul 3, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Our own analysis of that scan says the posts were off-subject. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Crw? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- AI-powered web search returning full page content, multi-engine and structured
- Scrape any URL into clean markdown, JSON, or HTML
- Full JS rendering for single-page applications
- Smart wait — waits for content to load before extracting
- Crawl an entire site with every page returned as structured data
- Map every URL on a domain without full page loads
- Extract structured JSON via prompt or JSON schema (managed LLM, paid plans)
- Parse web-hosted PDFs to markdown, CPU-only and billed per page with no LLM
- Monitor pages for changes with an AI verdict and a webhook
- Interactive scraping actions: click, scroll, write, wait, press
- Built-in anti-detection with header rotation and CAPTCHA handling, no proxy setup
- Selective caching with configurable caching patterns
- Source attribution — every response includes source URLs and metadata
- MCP server with one-command install: npx crw-mcp@latest install
- Self-host the AGPL-3.0 engine as a single ~8MB binary with no runtime dependencies
About Crw
fastCRW (CRW) is a web data API for AI agent developers who want search, scrape, crawl, map, extract, and monitor behind a single REST endpoint instead of stitching together four vendors. You call it from Python (`pip install crw`), Node.js (`npm install crw-sdk`), cURL, or MCP, and `npx crw-mcp@latest install` auto-detects your agent, then wires in the skill and the MCP server with no config file to hand-edit. Responses come back as LLM-ready markdown, JSON, or HTML with source URLs and metadata attached, so your users can verify an answer and your team can debug it. Scraping handles the annoying parts in-house: full JS rendering, smart wait for late-loading content, interactive actions (click, scroll, write, wait, press), header rotation and CAPTCHA handling with no proxy setup, and selective caching you control by pattern. Parse turns web-hosted PDFs into clean markdown CPU-only with no LLM, while Extract pulls structured JSON from a prompt or JSON schema using a managed LLM that is only available on paid plans and metered by usage. Metering is flat and easy to model: 1 credit per page on scrape, crawl, and map regardless of renderer, 2 credits per search of 10 results, and failed requests are never billed. The entry point is genuinely free — 500 credits to start plus up to 500 more from onboarding tasks, no card required — and paid usage-based plans start at Hobby $11/mo billed $132/yr. The engine itself is AGPL-3.0 and ships as a single ~8MB binary with no runtime dependencies, so you can self-host at zero license cost and point any client at your own instance with CRW_BASE_URL. Cloud and self-host expose identical scrape, crawl, map, and search endpoints. Compared with Firecrawl, Tavily, Exa, and Parallel.ai, fastCRW competes on published recall numbers and auditability rather than dashboard polish — it publishes 63.74% truth-recall on Firecrawl's own 1,000-URL dataset against Firecrawl's 56.04%, and 61.0% research recall on the ArXivQA benchmark against
Behind the Verdict
The thing that separates fastCRW from the rest of the scrape-API crowd is not the feature list — everyone has scrape, crawl, and search now. It's that you can read the exact code they run in the cloud, under AGPL-3.0, as a single ~8MB binary with no runtime dependencies, and point any client at it with one environment variable. In practice that matters most for teams whose procurement or security review ends at "show me the crawler" rather than "send me your SOC 2." We'd reach for this when an agent needs web access wired in fast. `npx crw-mcp@latest install` detects Claude Code, Cursor, Windsurf, Cline, Codex, and the rest, and drops in both the skill and the MCP server without config. If you're already inside n8n, Make, or Zapier, the API path is equally short — one key, one endpoint, markdown back. The credit model is the other quiet advantage. 1 credit per page on scrape, crawl, and map no matter which renderer fires, 2 credits per search, failed requests never billed. That's rare in a category where proxy and browser-rendering surcharges quietly double the bill. Hobby's 5,000 monthly credits is roughly 5,000 page scrapes, and 100,000 credits on Standard at $59/mo works out to about $0.00069 a page, or $2.30 a day. Where it bites: Extract is metered dynamically by token usage and is paid-plans-only, so a high-volume structured extraction pipeline can't be modeled from the per-page rate alone — do the token math before you commit. And there's an active capacity rollout in progress. Sustained concurrent throughput may run below each plan's provisioned concurrency target, and requests over live capacity get a 503 with a Retry-After header. If your workload is bursty or latency-critical, build retry logic in from day one rather than discovering it in
Researching Crw? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Crw actually fits — and what changes day-one when you adopt it.
You install the SDK, run `npx crw-mcp@latest install` to wire the skill and MCP server into Claude Code or Cursor, then replace your ad-hoc fetch-plus-parser with a single `crw.search()` call that returns titles, URLs, and snippets for your agent's tool loop.
Outcome: Web access lands behind one API key in an afternoon, and the 500 signup credits plus up to 500 from onboarding tasks cover your evaluation without a card.
You crawl a documentation site with `POST /v1/crawl` at maxPages 5, poll `GET /v1/crawl/{id}` for completion, and feed the returned markdown into your chunker, using Map first to confirm you have discovered every URL on the domain.
Outcome: A clean, attributable corpus at 1 credit per page — around $2.30 a day on Standard's published rate — with every response carrying source URLs and metadata for citations.
You pull the AGPL-3.0 engine as a single ~8MB binary, run it at localhost:3000 with no auth header, and set CRW_BASE_URL on your existing client so scrape, crawl, map, and search calls hit your instance instead of the cloud.
Outcome: No rewrite and no license cost, and scraped data never leaves your infrastructure; a commercial license is available if you later need a private deployment without copyleft.
Use Cases
- Scrape any URL into clean markdown for LLM ingestion or RAG pipelines.
- Search the web and retrieve full page content for agent tool calls.
- Crawl entire websites to build structured datasets for AI training.
- Map domain endpoints to discover all pages for migration or testing.
- Extract structured JSON from e-commerce or review pages using schema prompts.
- Monitor pages for meaningful changes and receive AI-generated verdicts.
- Parse PDF documents to markdown for document processing without an LLM.
- Self-host the engine to keep all web scraping data fully on your own infrastructure.
Limitations
- Extract uses a managed LLM metered by usage and is available on paid plans only.
- Concurrency is provisioned per plan (3 on Free up to 150 on Scale), and during the capacity rollout sustained throughput may run below published targets with requests over live capacity receiving a 503 and Retry-After header.
- The Free tier is 500 starting credits (up to 500 more from onboarding tasks), not a monthly allowance, and storage/rendering features such as screenshot output return HTTP 422 (an acknowledged gap).
- Self-hosting the AGPL-3.0 engine means running and maintaining your own infrastructure.
as of 2026-10-08
Verification history
We have re-verified Crw 10 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 10 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Crw tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free
$0/mo
Ideal for
Developers evaluating fastCRW against Firecrawl or Tavily on a single project, with no card and no commitment.
What this tier adds
Starting tier — 500 credits one-time plus up to 500 more from onboarding tasks, 3 concurrent requests, Scrape/Crawl/Map/Search, no managed Extract.
Hobby
$11/mo (billed $132/yr)
Ideal for
Solo agent developer or side project scraping up to about 5,000 pages a month behind one API key.
What this tier adds
Adds 5,000 credits/month that reset monthly, 10 concurrent requests, MCP server integration, and managed LLM Extract, at $0.0026 per page.
Standard
$59/mo (billed $708/yr)
Ideal for
Small RAG or automation team running roughly 100,000 pages a month who needs more concurrency and a support channel.
What this tier adds
Jumps to 100,000 credits/month, 50 concurrent requests, $0.00069 per page, Standard support, and top-up credits at $47 / 35k that never expire.
Growth
$237/mo (billed $2,844/yr)
Ideal for
Production agent platform at roughly 500,000 pages a month that needs headroom and fast support.
What this tier adds
Adds 500,000 credits/month, 100 concurrent requests, $0.000558 per page, priority support, and auto-recharge on by default with an optional spending cap.
Scale
$467/mo (billed $5,604/yr)
Ideal for
High-volume pipeline at roughly 1,000,000 pages a month where per-page cost has flattened out.
What this tier adds
Adds 1,000,000 credits/month, 150 concurrent requests, $0.000549 per page, and top-up credits at $299 / 500k that never expire.
Enterprise
Custom
Ideal for
Organizations past a million pages a month or with data-residency, SLA, or dedicated-infrastructure requirements.
What this tier adds
Adds custom credit volume, dedicated clusters, custom concurrency, premier support with SLA, and a commercial license option for private self-hosted deployment without copyleft.
Where the pricing makes sense
The company stage and team size where Crw's pricing actually pencils out — and where peers do it cheaper.
Published plans run Free, then Hobby at $11/mo billed $132/yr, Standard at $59/mo billed $708/yr, Growth at $237/mo billed $2,844/yr, and Scale at $467/mo billed $5,604/yr, with Enterprise custom. At roughly $0.00069 per page on Standard you are priced below Firecrawl's managed tiers for comparable scrape volume, and Standard's 100,000 credits a month covers most RAG builds; solo developers should start on Hobby, while teams past a million pages a month should price Scale against self-hosting
Setup time & first value
How long it actually takes to get something useful out of Crw — broken out by persona, not the marketing-page minute.
For an individual developer, first value is minutes: `pip install crw` or `npm install crw-sdk`, drop in an API key, and the first search or scrape returns in one call, with `npx crw-mcp@latest install` auto-wiring the MCP server into a supported agent. For a team, budget roughly half a day to decide between cloud and self-host and wire keys into CI. Self-hosting starts with the ~8MB binary or
Switching to or from Crw
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Firecrawl: fastCRW publishes a Firecrawl-compatible surface and reports 63.74% truth-recall versus Firecrawl's 56.04% on Firecrawl's own 1,000-URL dataset, so you can benchmark before cutting over.
- →From Tavily: swap the SDK for `crw` on PyPI or `crw-sdk` on npm, point at /v1/search, and keep the same agent tool call shape.
- →From Parallel.ai: fastCRW publishes a surface-by-surface Parallel.ai comparison covering search, extract, and monitor endpoints to map each call.
- →From Crawl4AI: fastCRW publishes a dedicated alternatives comparison and exposes the same crawl, map, and scrape verbs behind one hosted or self-hosted endpoint.
- →From a self-built scraper: keep your parsing code by switching to the hosted API, or run the AGPL-3.0 binary locally and set CRW_BASE_URL.
- ↗To Firecrawl: move to a GUI-first workflow with a managed dashboard if your team needs visual job management over a REST API and MCP server.
- ↗To self-hosting fastCRW: not a vendor migration — deploy the same AGPL-3.0 binary and set CRW_BASE_URL, leaving the scrape, crawl, map, and search endpoints unchanged.
- ↗To Parallel.ai or Exa: switch if your workload centers on their particular retrieval or task-API surface rather than per-page scrape credits.
- ↗To a self-built crawler: rebuild the anti-bot, JS-rendering, and caching layer in-house if you need full control and can absorb the maintenance.
Integrations
Resources & Guides
- Documentationfastcrw.com
Docs · Crw
Full product docs from fastcrw.com
- Resourcedocs.fastcrw.com
Home · Crw
Helpful link from docs.fastcrw.com
- Resourcefastcrw.com
Playground · Crw
Helpful link from fastcrw.com
- Resourcefastcrw.com
Integrations · Crw
Helpful link from fastcrw.com
- Resourcefastcrw.com
Benchmarks · Crw
Helpful link from fastcrw.com
- Resourcefastcrw.com
Alternatives · Crw
Helpful link from fastcrw.com
- Resourcefastcrw.com
Blog · Crw
Helpful link from fastcrw.com
- Resourcegithub.com
Crw · Crw
Helpful link from github.com
Tutorials & Learning
YouTube returned 6 videos for “Crw”, and we withheld 6: 6 could not be judged, because “Crw” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Crw.
Official links
Tools that pair well with Crw
Common stack mates teams adopt alongside Crw, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Crw vs Spider Cloud
Choose Crw if you want an open-source, drop-in Firecrawl replacement with tiny self-host footprint and a proven 63.74% truth-recall rate. Choose Spider Cloud if you need a high-success (99.9%) cloud API with AI Studio, Browser AI commands, and 1,000+ ready-made scrapers. For budget-conscious, open-source-first teams, Crw wins; for convenience and anti-bot muscle, Spider Cloud leads.
Crw vs Presto Voice
Presto Voice and Crw serve completely different domains. Presto Voice is purpose-built for QSR drive-thru automation, delivering proven revenue lift and high accuracy. Crw is an open-source web data API for AI agents, ideal for developers needing low-latency scraping. Choose based on your industry: restaurant operations or AI data pipeline.
Crw vs Temporal Ai
Choose Temporal AI if you need durable, fault-tolerant orchestration for multi-step AI agents or microservices with retries and human-in-the-loop. Choose Crw (fastCRW) if you need a fast, open-source web scraping API that drops into Firecrawl/Tavily workflows. They are complementary: Temporal orchestrates Crw as an activity.
Alternatives to Crw
View all
Popular in Web Scraping & Search APIs
Spider Cloud
Spider Cloud is a web scraping, crawling, and search API that streams live pages as markdown or JSON for AI agents and RAG pipelines.
Frequently Asked Questions
Categories
Used Crw? Help shape our editorial sentiment research.