50–100 KB
a typical markdown scrape
web data infrastructureClean web data at scale
Markdown, raw HTML, or thousands of URLs in bulk — and a webhook with the new content when a page changes. Pay for bytes used, not requests made.
curl -X POST https://api.bytekit.com/v1/scrape \ -H "Authorization: Bearer $BYTEKIT_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "url": "https://bytekit.com", "formats": ["markdown"] }'
Response
{ }
No response yet
Click Run to send the request
what you can doOne API, four endpoints
/scrape
Page content as clean markdown for agents and RAG ingestion. Most pages return in 1–2 seconds.
/monitors
Watch any page. When it changes, your webhook gets the new content.
/fetch
Raw HTTP fetch — no JS render, for APIs and static pages.
/bulk
Submit many URLs; results delivered by webhook as each completes.
what's includedThe hard parts are handled
Stealth fingerprints. Mobile and residential proxies in 100+ countries. CAPTCHA solving. Works on sites protected by DataDome, Cloudflare, Akamai, PerimeterX, and the rest.
All included on every plan. No tier upgrades. No add-on fees. Protected sites cost the same as any other capture.
You worry about your product.
We worry about bot detection.
comparePick the tool that fits the job
| Approach | Good for | Tradeoff |
|---|---|---|
| DIY Playwright | Full control | You own infra, retries, proxies, storage |
| Browser-only APIs | Dynamic pages | More expensive for simple pages |
| No-code scrapers | Manual workflows | Harder to embed in production systems |
| ByteKit | Agents, RAG, monitoring, captures | Public pages only — no logged-in workflows |
pricingPay for bytes used, not requests made
A typical markdown scrape is 50–100 KB. 1 GB covers roughly 10,000–20,000 scrapes. Cache hits cost half. Failed captures cost zero.
Plans from $29/mo, or pay-as-you-go at $8/GB. Expiring credits suck, so unused balance rolls over every month.
10–20k
scrapes per 1 GB
0.5x
cache hits cost half
0
failed captures are free
Test your own URLs with 50 MB free.
Built for agents, RAG, monitoring, and captures. Not for logged-in workflows, walled-garden social, or point-and-click scraping.