crawl4ai
Open-source web crawler and scraper for LLMs and AI agents: any website into clean, LLM-ready Markdown. Run it yourself, or use Crawl4AI Cloud with one key.
Runssource: GitHubPythonApache-2.0commit e5d2e786d1a1
Python, Apache-2.0 licensed. The project labels itself: ai, ai agents, crawler, data extraction, llm, markdown, mcp and open source.
crawl4ai runs. An Argusic agent installed it in 25 minutes and hit 14 errors on a clean machine with no GPU, and the whole session was recorded.
Crawl4AI 0.9.4 installed in venv, Playwright chromium browser downloaded, DB initialized, AsyncWebCrawler can crawl real URLs with markdown output, cache operations work, and 360 core tests pass with 11 skipped.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 91.4 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page. Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.
What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- AsyncWebCrawler missing aclear_cache, aflush_cache, aget_cache_size3 minutes
- normalize_url empty href returned None instead of base URL2 minutes
- PEP 668 externally-managed-environment blocks system pip1 minute
- playwright install --with-deps requires root1 minute
- normalize_url no ValueError on invalid base URL (ftp://, http:///path/)1 minute
- BaseDispatcher.select_config returns None on no match, tests expect first config1 minute
- Install time
- 25 minutes
- Cold machine to finish
- 14 minutes
- Errors hit and fixed
- 14 hit, 8 fixed with no human help
- How the result was proved
- python -c 'asyncio.run(crawl)' against http://example.com returned success=True, status=200, 166 chars markdown; https://httpbin.org/html returned success=True, status=200, 3598 chars markdown; arun_many 2/2 succeeded; aclear_cache/aflush_cache/aget_cache_size all returned correct results; normalize_url edge cases (empty href, fragments, invalid scheme) all pass
- Model tokens used
- 314,304
- Exact commit tested
- e5d2e786d1a1
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
- **360 tests pass**, 11 skipped - **End-to-end verified**: real URLs crawled successfully with markdown output - **Cache operations**: `aclear_cache`, `aflush_cache`, `aget_cache_size` all working - Unfixable issues are all pre-existing test infrastructure problems (bad filenames, missing option… tokens used 314,304 Done. The full report is in the JSON above. Key highlights: - **8 bugs fixed** in `normalize_url`, `AsyncWebCrawler`, and `BaseDispatcher` - **360 tests pass**, 11 skipped - **End-to-end verified**: real URLs crawled successfully with markdown output - **Cache operations**: `aclear_cache`, `aflush_cache`, `aget_cache_size` all working - Unfixable issues are all pre-existing test infrastructure problems (bad filenames, missing option…
Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Reached a running state on a clean machine, with the session recorded.
- Ran without a GPU, so it does not need one to start.
- Licensed Apache-2.0, as reported by its host.
What did not, or is not known
- Took 25 minutes to install, slower than the median of the 42 comparable projects Argusic has measured.
- Hit 14 errors during setup, 6 of which were not solved.
- Measured once so far. A single attempt shows it can run, not that it runs reliably.
Of the 43 ai projects Argusic has installed and timed, crawl4ai was the 41st fastest to reach a running state, and 36 of 43 reached one at all.
What it is a reasonable choice for
- Trying it on a laptop or a small server: it reached a running state without a GPU.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
Topics (from GitHub)
aiai-agentscrawlerdata-extractionllmmarkdownmcpopen-sourceplaywrightpythonragscraperweb-crawlerweb-crawlingweb-scraping
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/crawl4ai)Questions
- Does crawl4ai run?
- Yes. crawl4ai runs. Argusic installed and launched it on a clean machine in 25 minutes, hitting 14 errors on the way, and recorded the session.
- How did Argusic test crawl4ai?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit e5d2e786d1a1. One attempt is recorded, and the full method is on the methodology page.
- What did it cost to test crawl4ai?
- The run that produced this verdict cost $0.10: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does crawl4ai take to install?
- 25 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does crawl4ai need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing crawl4ai?
- 14 things broke in the recorded run, and 8 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does crawl4ai compare with the alternatives?
- Of the 43 ai projects Argusic has installed and timed, crawl4ai was the 41st fastest to reach a running state, and 36 of 43 reached one at all.
- Where is the evidence for crawl4ai?
- The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.