What 1,000 Scraped Pages Actually Cost in Q1 2026
Pricing across the scraping stack now sits in three distinct tiers: bandwidth proxies (per-GB), unblocker APIs (per-request), and managed scrapers (per-result). 1,000 fetches of a 50KB JS-rendered page costs anywhere from $0.30 to $10 depending on tier — a 30× spread.
The scraping infrastructure market does not publish a comparable per-page cost figure. Each vendor uses its own pricing primitive (per-GB, per-request, per-result, per-credit, per-month) and resists normalization. Most public comparisons are either marketing-driven or compare incompatible tiers.
Here is the actual math, normalized to one common target: 1,000 successful fetches of a 50KB JavaScript-rendered page, with anti-bot bypass where required, on the published rate cards as of Q1 2026.
The spread is roughly 30x between the cheapest tier and the most expensive. That is the headline.
Tier 1: bandwidth proxies (BYO scraping logic)
These vendors sell IP rotation. The customer brings the scraper. The cost per 1,000 pages depends on how much HTML and how many supporting assets the page actually transfers.
| Vendor | Rate | Estimated $/1k pages (50KB JS) |
|---|---|---|
| Bright Data residential PAYG | $0.60/GB | ~$0.03 |
| Bright Data residential standard | $5.88–$10.50/GB | ~$0.30–$0.53 |
| Oxylabs residential premium | $10/GB | ~$0.50 |
| Decodo (Smartproxy) residential | comparable to Oxylabs | ~$0.30–$0.50 |
The 50KB headline is conservative. A modern JS-rendered page typically pulls 200KB+ of supporting assets (frameworks, fonts, images), pushing actual bandwidth toward 250KB per page. The real cost on most JS-heavy targets is 5x the table number.
The bandwidth tier is structurally the cheapest, but the customer pays separately for compute (running the headless browser), proxy concurrency, and any anti-bot bypass that requires more than a clean residential IP.
Tier 2: unblocker APIs (vendor handles JS + anti-bot)
These vendors sell the outcome: a clean rendered page, regardless of anti-bot complexity. The customer sends a URL, gets back HTML or markdown. Pricing is per successful request.
| Vendor | Rate | $/1k pages |
|---|---|---|
| Bright Data Web Unlocker | $3 per 1,000 | $3.00 |
| ZenRows | $69/mo for 10k results | ~$6.90 |
| ScrapingBee | $49/mo (Freelance) | varies, ~$1–$3 typical |
| Oxylabs Web Unblocker | $9.40/GB equivalent | ~$5–$10 |
The unblocker tier is roughly 10x the bandwidth tier on equivalent successful pages, but the customer no longer pays separately for compute, JS rendering, or basic anti-bot. For mid-volume operations, the all-in cost is competitive with running the bandwidth tier in-house once engineering time is included.
Tier 3: managed scrapers (vendor handles target-specific extraction)
These vendors sell pre-built scrapers for known targets — Apify Actors, ScraperAPI’s e-commerce endpoints, Bright Data’s dataset products. Pricing is per result, structured.
| Vendor | Rate | $/1k results |
|---|---|---|
| ScraperAPI e-commerce | $299/mo for 600k requests | ~$0.50 |
| Apify PPE actors | $0.0015–$0.01 per result | $1.50–$10 |
| Firecrawl extraction | $0.004 per page (5x credit multiplier) | $4.00 |
| Bright Data datasets (one-time) | $250 per 100k records | $2.50 |
| Bright Data datasets (monthly refresh) | up to 80% off | as low as $0.50 |
The managed tier pricing range is 20x — from $0.50 to $10 per 1,000 results. That spread reflects (a) target popularity (common targets are cheaper because vendors compete), (b) freshness requirements (real-time costs more than refresh-on-monthly), and (c) extraction complexity (a single product field is cheaper than a full structured record).
What this means for the buyer
For a buyer evaluating where on the stack to operate, the math suggests:
Volume above 100M pages/month: bandwidth tier wins. The fixed cost of building and maintaining the scraping stack is amortized across enough volume that engineering time becomes a small fraction of unit cost.
Volume between 1M and 100M pages/month: unblocker tier usually wins. The all-in cost (vendor + minimal engineering) beats DIY bandwidth + scraping infrastructure for most teams.
Volume below 1M pages/month, or specific targets: managed tier wins. The marginal cost of buying a finished result from a vendor that has already solved the target-specific challenges is lower than building any of it yourself.
Common, high-volume targets (Amazon products, LinkedIn profiles, Indeed jobs): vendor datasets win on price, but only if the freshness requirements match the vendor’s refresh cadence. For real-time, the managed tier with PPE pricing wins.
The 30x spread between the cheapest and most expensive published tier is real, but it is not arbitrary. It reflects what the buyer actually buys: progressively more of the engineering and infrastructure problem solved on their behalf.
What the Apify Store ecosystem looks like in this frame
The Q1 2026 censuses on this site show that the Apify Store sits in tier 3 (managed scrapers), with most actors priced between $0.0015 and $0.01 per result. That positions the Store as a competitor to Bright Data datasets on per-record pricing, and as a competitor to ScraperAPI on target coverage.
The open publisher-side question is whether the Store’s pricing range — which currently averages closer to $0.005 per result — should compress toward Bright Data’s $0.0025/record published rate. The compression argument: as Bright Data publishes more dataset packages on more targets, every Apify Actor on those targets faces direct price competition. The non-compression argument: Bright Data’s dataset products are fundamentally batch (one-time pull or monthly refresh), while Apify Actors are real-time. Buyers who need fresh data will pay the premium.
The macro pattern across all three tiers is the same one visible in Clay’s $3.1bn valuation and the e-commerce data floor: collection is commoditizing, margin is moving to the orchestration / freshness / convenience layer. Vendors who win the next phase will be the ones whose pricing is closest to what the buyer actually needs to pay.
The 30x spread will not last forever. The bottom of the spread is at the floor of the marginal cost. The top of the spread is sustained by accumulated brand and engineering convenience. Both will compress over the next twelve to eighteen months, and the convergence point will be visible in the next quarterly census.
Sources