2026-07-16 · 9 min read
Web scraping in 2026 costs anywhere from $0.13 per 1,000 requests to six figures a year, depending on who does the work. Scraping APIs bill usage and assume you bring engineers. No-code tools run $49 to $249 a month and assume you bring patience. AI agent platforms like WebRobot run $79 to $699 flat. Managed services start around $199 a month, and an in-house build costs whatever your engineers cost, forever. Here are the real numbers, from vendors' own public pricing pages.
Last updated July 2026 · prices verified against vendor pricing pages
Robot console · WR-01
Standing by Running · s Complete · rows
1 · Pick a target
3 · Fields to extract
Agent log
Crawl graph
Extracted data · rows
Want this data fresh every morning, without lifting a finger?
| Approach | Typical price | Who operates it | The cost not on the pricing page |
|---|---|---|---|
| Scraping APIs Zyte, Bright Data, Apify |
Zyte from ~$0.13/1K simple requests; Bright Data $1.50/1K records or $499/mo for 384K; Apify from $29/mo in credits | Your developers | The parser, storage, scheduling, and repair code are all yours. Browser rendering costs 8 to 12x a plain fetch |
| No-code tools Octoparse, Browse AI, ParseHub |
Free demo tiers; paid from $48 to $89/mo entry, $249/mo mid-tier | You, clicking | Selector workflows break on redesigns; rebuilding them is your recurring, unbudgeted job |
| AI agent platforms WebRobot |
Flat $79/mo (Launch), $249/mo (Scale), $699/mo (Autopilot); yearly billing $63/$199/$559 | Anyone, in plain English | No free tier; record caps per plan (10K/100K/1M per month) |
| Managed services ScrapeHero and similar |
From ~$199/mo subscriptions; $550+/mo on-demand; enterprise to six figures/yr | The vendor's team | Slow iteration: every field change is a ticket to someone else's backlog |
| In-house build Python, Scrapy, proxies |
"Free" software; industry estimates put a maintained pipeline at $80,000 to $150,000/yr in engineering and infrastructure | Your engineers | All of it. Proxies, monitoring, on-call, and the redesign treadmill are permanent line items |
Two of those rows have honest fine print worth repeating: Zyte only bills successful responses, which is genuinely fair, and Bright Data's Web Scraper API includes a 5,000-record free monthly tier. We keep deeper teardowns of both: Zyte pricing and fit and Bright Data pricing and fit.
DRIVER 1
A static HTML page is nearly free to fetch. A JavaScript-heavy storefront needs full browser rendering, which on usage-priced APIs costs roughly 8 to 12 times more per request. Zyte formalizes this with five cost tiers per website; other vendors just bill the bandwidth.
DRIVER 2
1,000 rows once is pocket change everywhere. 100,000 rows refreshed hourly is where models diverge: usage pricing scales linearly with your ambition, flat plans cap it. Decide your refresh cadence before comparing vendors, or you are comparing different products.
DRIVER 3
Defended sites need residential proxies and retries; data behind logins and search forms needs an agent that can click and type. On DIY stacks these are separate subscriptions. On WebRobot, agent actions are a plan feature, not a meter.
DRIVER 4
The quiet majority of total cost. Every target redesign breaks selector-based extraction, and someone pays to fix it: your engineer, your ops person, or a vendor whose robot self-heals. Ask every vendor one question: "what happens when the site changes?"
WORKLOAD A
Scraping API: a few dollars in credits plus the developer you do not have. No-code tool: $48 to $89 and your Saturday rebuilding workflows twice a year. WebRobot Launch: $79 flat ($63 yearly), described in a sentence. Managed service: overkill at $199+.
WORKLOAD B
Roughly 290,000 checks a month against rendered retail pages. Usage APIs: hundreds of dollars of browser-tier requests, plus your alerting code. WebRobot Scale: $249 flat ($199 yearly) including hourly runs and Slack alerts. This is the workload flat pricing was built for.
WORKLOAD C
Honest answer: usage-priced infrastructure wins. At $0.06 to $0.13 per 1,000 simple requests on commitment tiers, Zyte-class APIs deliver this for hundreds of dollars, if you have the engineers. Flat-priced agents are not the tool for raw crawling at this scale.
A budgeting note: scraping spend has a habit of fragmenting into proxy subscriptions, API credits, and tool seats across teams, and finance only sees the total when it is embarrassing. If SaaS line items are already sprawling, a cloud cost management tool will surface what you are actually paying for data collection across the company before you negotiate any single vendor.
WHAT THE SPRINT BUYS
A Scrapy spider that works on today's version of the target site, on your laptop, at low volume. Genuinely fine for a one-off research pull, and free apart from the afternoon it took. If that is the whole job, build it and move on.
WHAT PRODUCTION ADDS
Proxies ($50 to $500+/mo), hosting, monitoring, alerting, retry logic, a queue, and the redesign treadmill: every target ships a new front end on its own schedule, and your pipeline breaks on each one. Industry estimates for a maintained pipeline run $80,000 to $150,000 a year. That is 27 to 50 years of WebRobot Scale.
The decision rule that holds up: build when scraping is your product or your volume is in the millions of pages; buy when scraping is an input to your actual job. The full comparison of approaches is in web scraping vs API, and the tool landscape in best web scraping tools in 2026.
FINAL ASSEMBLY
Describe the data you need in plain English. WebRobot extracts and delivers it on a schedule, from $79 per month flat.