Crawl4AI compatible

The call, before and after

# Before
curl -X POST http://localhost:11235/crawl \
  -H "Authorization: Bearer $CRAWL4AI_API_TOKEN" -H "Content-Type: application/json" \
  -d '{"urls": ["https://en.wikipedia.org/wiki/Web_scraping"]}'

# After
curl -X POST https://api.specstream.ai/crawl4ai/crawl \
  -H "Authorization: Bearer $SPECSTREAM_API_KEY" -H "Content-Type: application/json" \
  -d '{"urls": ["https://en.wikipedia.org/wiki/Web_scraping"]}'

The reply is Crawl4AI's: {"success": true, "results": [{"url": "...", "success": true, "status_code": 200, "markdown": {"raw_markdown": "# Web scraping ..."}, "links": {"internal": [...], "external": [...]}, ...}], "server_processing_time_s": ...}.

With Crawl4AI's Python client, crawl4ai 0.9.5, the switch is two steps: build Crawl4aiDockerClient(base_url="https://api.specstream.ai"), then call await client.authenticate(KEY) with your SpecStream key (the client names that parameter email). client.crawl([...]) then runs as before. In your own HTTP call it is the address and the key, and the bodies the self-hosting guide documents run unchanged.

What the route reads, and what it refuses

/crawl, applied: urls, 1 to 100 in one call. Each page answers on its own, so of three URLs with one missing, two come back with success true and the third with success false. A page the site answers 404 comes back with status_code 404 and error_message set.

/md, applied: url and f as raw. The reply echoes url, filter, query and cache and gives the page as markdown.

Refused, in Crawl4AI's own {"detail": ...} body: an empty urls or more than 100 gets 400, f that is not a filter gets 400, a URL with no scheme gets 400, an /md of a page with no text gets 500 and is not charged, and a wrong key gets 401. Every parameter was measured on production on Nov 10, 2026, and the full pass table lists each one and its result.

Every field that comes back

/crawl returns success, server_processing_time_s and results[], each with url, html, success, markdown.raw_markdown, links.internal, links.external, metadata, status_code and error_message. /md returns url, filter, query, cache, markdown and success.

The route does not return cleaned_html, markdown.fit_markdown, media, screenshot or pdf on a result, or server_memory_delta_mb and server_peak_memory_mb on a crawl.

One price a page, and no server

Crawl4AI's library is free and the server runs on your own hosting, with its browser and memory to size and keep up. On SpecStream every page that gives text costs $0.0008, which is $0.80 per 1,000, a page with no text costs nothing, and a busy call gets 429 with Retry-After and charges nothing. Credit you buy never expires.

Scored on the official benchmarks

The SpecStream Extract API scores 0.908 F1 on WCXB. The Search API scores 97.9 F-score on all 1,000 SimpleQA Verified questions, 978 of them correct, and 64.0 item F1 and 43.4 row F1 on all 200 WideSearch tasks. Every score and its method is on the benchmarks page.

You decide what is kept, on each key

Create the key with zero_retention and a call made with it keeps no request text and no answer after the reply. spend_cap_cents is the most the key can spend.

curl -X POST https://api.specstream.ai/v1/account/keys \
  -H "Authorization: Bearer $SPECSTREAM_API_KEY" -H "Content-Type: application/json" \
  -d '{"name": "production", "scopes": ["account:read", "finds:spend"], "spend_cap_cents": 500000, "zero_retention": true}'

Pages, search and rows on one key

The same key reads up to 20 pages or documents a call with the Extract API, runs site maps and crawls that follow links at https://api.specstream.ai/firecrawl, and searches the web at $5 per 1,000 searches. Deep Search returns every match for a request in plain words as structured rows, at 5¢, 15¢ or $1 a row, with a source URL on every value. The full comparison is on the Crawl4AI alternative page.

Crawl4AI's hosted API

A different API, with /scrape and /extract, on monthly plans from $10 a month, where the month's credits do not carry over. SpecStream answers the server's own /crawl and /md, so your code stays as it is.

Firecrawl

Scrape, map and crawl in credits: Standard is $99 a month for 100,000 credits. SpecStream answers Firecrawl's v2 calls at https://api.specstream.ai/firecrawl at $0.80 per 1,000 pages.

Spider

A scrape and crawl API billed by bandwidth and CPU time. SpecStream answers Spider's scrape and search at https://api.specstream.ai/spider, and a page costs $0.0008.

Apify

A platform for scrapers, with plans from $19 a month for $19 of usage. SpecStream reads any page at $0.0008 with no plan.