Skip to main content
Guides / Spider pricing and plans: what you pay and why
The full Spider logo.

Spider pricing and plans: what you pay and why

A breakdown of Spider's pay-as-you-go pricing, per-endpoint costs, volume discounts, and how billing works for web scraping and browser automation.

• 6 min read • Jeff Mendez •

Spider uses a pay-as-you-go model. You buy credits, spend them on requests, and they never expire. Pay as you go has no monthly subscription, no seat limits, and no hidden fees. Spider also offers Unlimited, a monthly concurrency plan, and AI Studio add ons. Both are optional and separate from your credit balance.

This guide covers how billing works, what each endpoint costs, and how to estimate your spend.

How billing works

Every API request is billed on two components:

  • Bandwidth: $1 per GB of content transferred (measured on pre-transformation bytes)
  • Compute: $0.0001 per minute of server-side processing time, one credit a minute, pro-rated to the millisecond

An attempt bills for the bandwidth and compute it used, whatever status the page returns. A failed page is not free. A blocked or empty page pays the 0.01 credit page minimum plus the compute it used, usually a fraction of a cent. Spider does not bill bandwidth for bytes it never received.

Credits are purchased in US dollars at a rate of 10,000 credits per $1. The smallest purchase is $5.

How minutes are metered

The meter is the same for the scraping API and the cloud browser, and these are the constants the billing code compiles in.

  • $1 buys 10,000 credits, so one credit is $0.0001.
  • Bandwidth is bytes transferred, divided by 1,000,000,000, times 10,000 credits. That is $1 per decimal gigabyte, counted on the bytes before any transformation.
  • Compute is elapsed milliseconds divided by 60,000. One credit per minute, pro-rated to the millisecond, with no rounding up to the next minute. A 12 second request is 0.2 credits.
  • An API request’s compute stops at 30 credits ($0.003), or at 50 credits ($0.005) when the request needed a headless browser.

A browser session bills the same two terms every 3 seconds while it is open, from init() to close(), with a last tick at close for the remainder. Recording a session multiplies the compute term by 1.15 and leaves bandwidth alone. A single tick’s compute stops at 50 credits. About every 30 seconds the server checks your balance, and at zero it ends the session.

Model calls made by act(), extract() and agent() are billed by your model provider, not by Spider.

If you meter a session yourself and your numbers disagree with your invoice, that is a bug. Tell us.

Per-endpoint cost estimates

These are average costs per request. Actual cost depends on page size and rendering time.

EndpointDescriptionAvg cost
/crawlCrawl and extract content across pages~$0.0003/req
/scrapeFetch and extract a single page~$0.0001/req
/linksCollect hyperlinks from a page~$0.0002/req
/screenshotCapture a page screenshot~$0.0006/req
/searchSearch across multiple engines~$0.001/req
/transformClean and normalize raw data$0.0001/obj
Proxy Mode (proxy.spider.cloud)Route through residential proxies~$1-4/GB
Cloud browser (browser.spider.cloud)Full CDP/WebDriver browser sessions$0.0001/min + $1/GB

Request modes and their cost impact

Spider supports three request modes. The mode you choose affects compute time, and forcing a browser adds bytes.

ModeCostHow it works
HTTP$0.016 per 100 light static pages, measured November 17, 2026Static fetching, no JavaScript.
SmartThe same as HTTP on pages that need no rendering, measured in the same runHTTP by default, moves to JS rendering when the page needs it.
BrowserNot measured here. A browser also fetches the scripts and data calls a page pulls in, and those bytes are on the meterFull browser rendering for JavaScript-heavy sites.

For most scraping workloads, smart mode gives the best balance between cost and reliability. Use http when you know the target serves static HTML. Use browser when you need full JavaScript execution on every page.

Set the mode with the request parameter:

import requests, os

headers = {
    'Authorization': f'Bearer {os.getenv("SPIDER_API_KEY")}',
    'Content-Type': 'application/json',
}

response = requests.post('https://api.spider.cloud/crawl',
  headers=headers,
  json={
    "url": "https://example.com",
    "limit": 25,
    "return_format": "markdown",
    "request": "smart"
  }
)

Add-ons

Some features add a small per-request cost on top of the base billing:

Add-onCostNotes
JS RenderingIncludedNo surcharge. Increases compute time, billed at the base compute rate.
Metadata extraction$0.0001/reqPage title, description, keywords, and other structured metadata.

Volume discounts

Larger pay-as-you-go top-ups add bonus credits automatically:

Purchase amountBonus creditsEffective discount
$500++5%5% more credits
$1,000++8%8% more credits
$2,000++12%12% more credits
$5,000++15%15% more credits

The top-up form shows $500 and $2,000 as presets. The $1,000 and $5,000 bonuses apply when you enter those amounts yourself.

Enterprise

Enterprise customers get custom SLAs, dedicated support, and volume pricing. Key capabilities include:

  • Unlimited concurrency
  • Global proxy network across 199+ countries
  • Smart browser escalation and the full native browser profile
  • Webhooks, streaming, and S3-compatible storage integration
  • Dedicated account management

Visit the enterprise page or email sales@spider.cloud to discuss your needs.

What pages cost, by tier

Cost follows page weight, so one number does not fit every site. The rows below say whether each figure is a measurement or a price rule. Each page kind was measured on 25 pages, fetched through /scrape with markdown output.

Page tierPer 1,000 pagesPer 100,000 pagesBasis
Light static pages, http or smart$0.16$16Measured November 17, 2026 on 25 catalogue pages of books.toscrape.com, a public practice site, with cache: false
Documentation and articles, smart$0.27 to $0.33$27 to $33Measured November 6, 2026, in the cost run
Store product pages and reference articles, smart$1.38 to $1.48$138 to $148Measured November 6, 2026, in the same run
Pages that need a premium proxyNot measuredNot measuredPrice rule: bandwidth through a proxy bills at 1.5x to 4x the $1 per GB rate

On the practice site, http and smart cost the same to the cent. The rates do not change with volume, so 10,000 pages come to $1.60 to $14.80 across the measured tiers and a million come to $160 to $1,480. Metadata extraction adds $0.0001 a request to any row, which is $0.10 per 1,000 pages and $100 per million.

These are small runs, and your pages will weigh more or less than ours did. Each response reports its own cost in costs.total_cost, so a 25 page sample of your own sites gives you the same table.

How Spider compares

Spider’s pay-as-you-go model avoids the subscription traps common in the scraping industry. If you would rather pay one monthly price for concurrency, Unlimited is there, and it is a choice, not a floor.

At the measured rates above, 10,000 pages cost $1.60 to $14.80, 100,000 cost $16 to $148 and a million cost $160 to $1,480, by page weight. Credits never expire. The comparison posts on the blog price each vendor against these tiers, each with its own date and sources.

Spider has no multiplier for JavaScript rendering or AI extraction. Two things do change the rate. Bandwidth that goes through a proxy bills at 1.5x to 4x the $1 per GB rate, depending on the proxy pool, and the compute term stays the same. The optional zero data retention setting bills every request at 2.5x on self-serve accounts. With other providers, enabling features like JS rendering or premium proxies can multiply the credit cost of each request by 5-75x.

Getting started

  1. Create an account (email or GitHub).
  2. Purchase credits starting at $5.
  3. Generate an API key.
  4. Start scraping.
curl -X POST https://api.spider.cloud/scrape \
  -H "Authorization: Bearer $SPIDER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com", "return_format": "markdown"}'

Credits never expire. Use them whenever you need them.

Run this on a page you care about

The playground sends the request this page describes and shows you the response. Keyless runs work without an account and are rate limited.