A flat $29/mo gets you 50k scrapes and unlimited concurrency. Or run the Go binary yourself for free. Pay-per-call ($0.0008/scrape) is in development — join the waitlist below.
For evaluation, side projects, and kicking the tires.
No subscription. No concurrency wall. Pay only for what you scrape. In development — join the waitlist.
Predictable monthly bill for steady recurring workloads.
High-volume API access, custom SLAs, or on-premise deployment — let's talk.
Free is for kicking the tires. Pay-as-you-go is for bursty agent workloads where usage is hard to predict. Pro is for steady, recurring monthly volume. All tiers share the same Apache 2.0 core — self-host whenever you want.
Pay-as-you-go fits production agents whose scrape volume swings between 0 and 50k on a given day — no wasted seat fee in quiet weeks.
Move to Pro once a month stabilizes above ≈35k scrapes and a flat bill becomes cheaper than $0.0008 per call.
Pro fits scheduled crawls, nightly doc ingests, and product catalogs where monthly volume is predictable.
Email us for a volume plan once you cross 500k scrapes/mo or need dedicated IPs.
Run the Apache 2.0 binary on your own VPS — no limits, no seat fee, every feature (including the extraction engine) included.
Switch to managed cloud only when you would rather not run a server yourself; start with Pay-as-you-go.
Pay-as-you-go is in active development. Target launch is the next product cycle. Join the waitlist from the pricing page and we'll email you when it's live — waitlist members get the first month of credits on the house. Today, your options are Free, Pro ($29/mo), or self-host.
Each URL you submit to the /v1/scrape endpoint counts as one scrape, regardless of the size of the page or the length of the output Markdown. Structured extraction via /v1/extract also counts as one scrape; the LLM call happens under your own API key (BYO LLM).
Because agent workloads are spiky. Most scraping SaaS forces you onto a monthly tier and caps the cheap tier at 5 concurrent requests — you end up either overpaying for idle capacity or getting throttled the moment your agent fleet wakes up. Our upcoming Pay-as-you-go tier is designed to remove both problems: no seat fee, no concurrency wall. Until it ships, Pro already removes the concurrency cap at a flat $29/mo.
Yes, fully. Purify is Apache 2.0 and the extraction engine, browser fallback, and MCP server all ship in the single Go binary. There is no closed-source module you have to pay the cloud to unlock. Self-hosted instances have no usage limits.
Purify does not have a two-tier codebase. Unlike some competitors that keep their anti-bot / rendering module closed-source (so the self-host option is a stripped-down version), Purify's self-host is the same binary that powers the managed cloud.
Model Context Protocol (MCP) is an open standard by Anthropic that lets AI agents like Claude and Cursor interact with external tools. Purify ships a built-in MCP server — drop one config file and your agent can scrape any URL.
Requests beyond the Free plan's 1,000/month return a 429. Top up $5 to switch to Pay-as-you-go, upgrade to Pro, or self-host the binary for no limits.
Cancel anytime from the dashboard. Access continues until the end of the current billing period. Pay-as-you-go has nothing to cancel — just stop topping up.
Yes. Purify tries HTTP first (~100ms) and only spins up a headless Chromium when a page actually needs JavaScript. You can force the browser path per request.
BYO-proxy support is on the near-term roadmap, not shipped yet. If this is a blocker for you, email hello@purify.verifly.pro with your use case — we prioritize the roadmap based on real requests.
All major credit cards via Paddle. Paddle handles global tax compliance so you don't have to.
Need more than 500,000 scrapes/month, dedicated IPs, or custom SLAs? Email hello@purify.verifly.pro.
Join developers building smarter AI agents — start free, no credit card needed.