Bright Data Pricing and Real Cost per 1,000 Pages, Plans and Per GB Rates

Bright Data is the largest vendor in this market and its pricing page reflects that. There are nine separate products with nine separate meters, some billed per request, some per gigabyte, some per IP address and some per record. None of them is priced in the unit you actually plan around, which is what it costs to collect a thousand pages.

This piece does that conversion. Every rate below was read directly from brightdata.com on 22 September 2026. Bright Data changes prices and runs rolling promotions, so check the live page before you commit a budget, and pay particular attention to which numbers are coupon prices with an expiry date.

The products and what each one meters

You are not choosing a plan so much as choosing a billing unit. This is the whole catalogue at its entry rate, as listed today.

Two of those lines decide almost every real bill: Web Unlocker if you want Bright Data to fetch the page, residential proxies if you want to run your own scraper through their network. The rest are either niche or sold on negotiated enterprise terms.

Web Unlocker is the easy number

Unlocker is the product most people mean when they compare Bright Data with a scraping API, and it is the only one already priced per request, so no conversion is needed.

The 499 USD tier is a genuine discount rather than a repackaged pay as you go: 499 divided by 383 gives 1.303 USD per 1,000, and additional requests stay at that rate instead of stepping back up. Unlocker bills only successful requests, which is the correct way to sell this and worth having.

Residential proxies are where the number moves

If you run your own crawler and only buy the network, you pay for bytes. That makes the cost of a page a property of the page, not of your plan, and it is the single biggest source of scraping budget surprises.

Those per GB figures are the monthly fee divided by the included gigabytes, which is the number that lands on your invoice. The per GB prices printed on the plan cards are rounded marketing values and do not reconcile exactly with the fee, so do the division yourself. Above 1 TB the price is negotiated.

Cost per 1,000 pages, by page weight

Now the conversion. A thousand pages at 150 KB each is 146.5 MB, or 0.1465 GB. At 500 KB it is 0.488 GB. At 1.5 MB, which is ordinary for a retail product page loaded with its images in a real browser, it is 1.465 GB. Multiply by the per GB rate and you have your answer.

Read that table twice, because it contains the one lesson that actually saves money on a per GB plan: blocking images, fonts and analytics scripts in your browser configuration is worth more than any discount tier. Going from a 1.5 MB render to a 500 KB one cuts the line by two thirds at every price point, and on most targets the data you want is in the HTML anyway.

It also explains why Unlocker often beats raw proxies for a team that has not tuned its crawler. At 1.5 MB a page you are paying 5.86 USD per 1,000 in bandwidth alone before you have written a parser. Unlocker is 1.50.

The coupon problem, and how to budget around it

Residential is advertised at 50 percent off with a coupon code, and the discount runs for three months. That is a real saving and there is nothing wrong with taking it. The mistake is building your unit economics on it.

If your model says a customer costs 0.59 USD per 1,000 pages to serve, and month four arrives at 1.17, your margin halves on a date you did not put in the calendar. Budget on the list rate, treat the coupon as a three month cash bonus, and you will not have to reprice anything. The same discipline applies to the annual discounts most vendors in this space offer.

The breakeven against flat per request pricing

Our own pricing is one rate per thousand successful requests regardless of target difficulty or page weight: 49 USD for 50,000 requests, 149 for 250,000, 499 for 1.5 million. That is 0.98, 0.60 and 0.33 USD per 1,000. Here is the comparison at three real volumes, using Unlocker as the closest equivalent product.

The gap widens with volume, which is the opposite of what most buyers expect from the larger vendor. It closes again at enterprise scale, where Bright Data negotiates rates that are not on any public page, so if you are buying several thousand dollars a month of anything, get a quote rather than trusting either this table or their pricing page.

Where Bright Data is the right buy

Plenty of cases, and pretending otherwise would waste your time. Their residential network is 400 million addresses across 195 countries, which is larger than anything we or most competitors operate, and if your work depends on obscure geographies that pool is the product. They sell ISP and datacenter proxies separately, so you can put cheap static addresses on easy targets and reserve residential for hard ones.

They also sell the answer rather than the tool. There are 800 or so prebuilt Scraper APIs for named sites at 0.75 USD per 1,000 records, and ready made datasets across 900 domains from 250 USD per 100,000 records. If the data already exists in their catalogue, buying it is cheaper and faster than collecting it, and you skip parser maintenance entirely.

For procurement, they are the safe choice: AWS Marketplace billing, account managers, premium SLA, SSO, a free tier to trial without a card. If you work somewhere that needs a vendor security review before a purchase order, that infrastructure has real value.

Where flat per request pricing wins

When you need to know the bill in advance. A per GB meter means your invoice is decided by how heavy the pages happened to be, how many retries the target forced, and whether someone shipped a crawler change that stopped blocking images. None of those are things a finance team can forecast, and all of them arrive after the money is spent.

Flat per request pricing removes the variable. Multiply pages by the rate and you have the number, whether the target is a static blog or sits behind Akamai Bot Manager, DataDome or Cloudflare. Refused attempts are not billed, so a hard target does not quietly cost you three times a soft one through retries.

It also wins on mixed workloads. If you are pulling ten thousand easy pages and two thousand heavily protected ones, credit and bandwidth models charge you a premium on exactly the pages you needed most, while a flat rate averages out. Once the rows are landing, moving them into a warehouse is a separate job, and a data integration platform that connects apps, APIs and databases is usually a better use of an afternoon than writing another loader by hand.

Questions buyers ask

How much does Bright Data cost per month? There is no single monthly price because every product meters differently. The realistic entry points today are pay as you go with no commitment, 499 USD a month for 383,000 Unlocker requests or 141 GB of residential traffic, 999 for 332 GB and 1,999 for 798 GB. A free tier of 5,000 requests exists on Unlocker, SERP API and Scraper APIs.

Is Bright Data expensive? At list rates, yes, compared with flat per request APIs, and the gap grows with volume: 1,951 USD against 499 for 1.5 million pages a month. At enterprise volumes with a negotiated contract that comparison stops holding, and if you need their geographic coverage or their ready made datasets, the premium buys something real.

Does Bright Data charge for failed requests? Web Unlocker bills only successful requests. The proxy networks bill bandwidth, which means a failed attempt still transfers bytes and still costs you, and on a heavily protected target retries can be a meaningful share of the bill.

What is the cheapest Bright Data plan? The free tier at zero, then pay as you go with no monthly commitment. If you are choosing purely on price for a volume you can predict, the per GB commitment tiers are cheaper per unit but lock up cash whether you use the allowance or not.

How do I compare Bright Data with other scraping vendors? Convert everything to cost per 1,000 pages first, using your own measured page weight rather than a vendor's example. We have done the same conversion for ScraperAPI, for Apify and for Oxylabs pricing per 1,000 pages, and all four sit next to our own flat plans in the same unit. Oxylabs is the closest direct comparison, since it publishes a migration guide for teams leaving Bright Data's Web Unlocker.