Bright Data pricing

About

Bright Data (formerly Luminati Networks) is a web-data platform that sells the full stack of “get public data off the web” tooling: proxy networks (residential, ISP/static-residential, datacenter, and mobile), unblocking and SERP APIs, a managed scraping browser, prebuilt and custom scraper APIs, ready-made datasets, and an eCommerce-focused insights product. It markets itself to 20,000+ customers worldwide and positions around the largest geographic proxy coverage (400M+ residential IPs across 195 countries).

The company has an unusually colorful lineage. It launched in 2014 as Luminati Networks, the B2B arm of the consumer Hola VPN — Hola’s “free VPN” resold its users’ idle bandwidth as residential exit nodes, a model a 296-point 2015 Hacker News thread put under a harsh spotlight. London private-equity fund EMK Capital acquired Luminati out of Hola in 2017 at roughly a $200M valuation, and the business rebranded to Bright Data in March 2021 (shedding the “Luminati”/“Illuminati” association). That history still shapes the product: residential and mobile networks sit behind a KYC compliance gate, and the company markets heavily on “ethical” and “transparent” data sourcing.

The buyer spectrum is wide — individual developers and small scraping shops buying a few GB or IPs pay-as-you-go, mid-market data teams committing to monthly plans, and enterprises negotiating custom volume contracts and managed data acquisition. Increasingly the center of gravity is AI: agent-oriented surfaces (Search & Extract, Agent Browser, an MCP server, a +50PB Web Archive, and LLM-ready data packages) now sit atop the core proxy and scraping infrastructure. That pivot is the growth engine — Bright Data reports crossing roughly $300M ARR in 2025 (up from about $128M in 2024) and targeting ~$400M by mid-2026, with more than 20,000 customers and a stated footprint across most of the top global LLM labs. It is also one of the few scraping vendors to have won in court: Meta’s contract claims were rejected at summary judgment in January 2024, and X Corp’s were dismissed as copyright-preempted in May 2024.

What ties the portfolio together is a consistently usage-based commercial model — but the meter changes per product. Bandwidth-heavy products (rotating residential, Browser API) bill per GB; static-IP products (ISP, datacenter) bill per dedicated IP; request-shaped products (Web Unlocker, SERP, Scraper API, Scraping Functions) bill per 1,000 successful results or page loads; and datasets bill per 1,000 records. This is the defining structural fact of Bright Data’s pricing and the reason a single headline number can’t describe it.

Pricing summary : one platform, four billing meters

Bright Data is usage-based across every product, but with four different value metrics depending on what the product consumes. Each product offers a pay-as-you-go entry point plus committed monthly tiers that lower the per-unit rate in exchange for a minimum monthly commitment and included volume; enterprise contracts are quoted.

• Per-GB (bandwidth): Rotating residential proxies ($8 → $2.50/GB) and the Browser API / Scraping Browser ($8 → $5/GB).
• Per-IP (dedicated static IPs): ISP proxies ($1.8 → $1.3/IP) and datacenter proxies ($1.40 → $0.90/IP), tiered by IP count (10 → 1,000).
• Per 1,000 requests/results/page-loads: Web Unlocker, SERP API, Web Scraper API, and Scraping Functions ($1.5 → $1.0 per 1k).
• Per 1,000 records: Datasets ($2.50/1k base, i.e. $250 per 100k) with refresh-cadence discounts up to 80%.
• Per-month subscription (by quote): Bright Insights eCommerce intelligence ($250–$400/mo to start).

What makes this different is the refusal to flatten everything onto one meter: Bright Data deliberately matches each product to the unit that best tracks its underlying cost (bandwidth for rotating IPs, dedicated IP count for static ones, successful requests for unblocking), which is a textbook example of picking the right value metric per product line. Because committed tiers add a seat-like monthly floor on top of metered overage, the overall shape is closer to a hybrid usage model than to pure pay-as-you-go. It also layers a compliance gate — residential and mobile networks require KYC before use — on top of an otherwise self-serve bandwidth-metered flow.

Pricing by product

The hero grid above summarizes every product line. The tables below give the full per-tier detail captured from each product’s pricing page on 2026-06-04. Where a vendor promo was live at capture, the post-discount price is noted; base rates are shown as the authoritative figure.

Proxy Networks — Residential (per GB)

A 50% promo (code RESIGB50) was live at capture, halving the above to $4 / $3.50 / $3 / $2.50 per GB for 3 months. Residential network use requires passing KYC.

Proxy Networks — ISP (static residential, per IP)

Proxy Networks — Datacenter (per IP)

Mobile proxies are referenced in the network FAQ (3G/4G/5G rotating IPs) but no public per-unit price is shown on the proxy-network page; mobile, like residential, requires KYC. Recorded as unknown pending a capturable surface.

Web Unlocker API & SERP API (per 1,000 results)

Scraping Functions (IDE, per 1,000 page loads)

Web Scraper API (per 1,000 records)

A 25% Scraper API promo (code APIS25) for 3 months was live at capture (e.g. $1 → $0.75/1k record on the platform overview).

Browser API / Scraping Browser (per GB)

Datasets (per 1,000 records)

Bright Insights (eCommerce intelligence, per month)

Sales motions across products: PLG / self-serve for all PAYG and committed monthly tiers (proxies, unlocker, SERP, scraper, functions, browser, datasets); sales-led for 1 TB+ proxy contracts, every product’s Enterprise tier, Bright Insights, and Managed Data Acquisition (from $1,500/mo per the platform overview).

Hidden costs : minimum commitments and KYC gating

The published per-unit rates are honest, but the real bill is shaped by three things the pricing cards underplay: the minimum monthly commitment on every discounted tier (it is a floor, not a rebate), the fact that bandwidth is counted on total bytes — request headers + payload + response headers + response data, not just the useful response, and the blended-meter problem of running several products at once. Two worked examples make the floor visible.

Archetype 1 — a mid-market scraping team. A data team commits to the Residential Business tier (to get $6/GB) and adds a Web Unlocker Growth plan for a separate JS-heavy target, then runs a monthly datasets pull on top.

The lesson: the two commitment floors alone are $1,498/mo before a single GB of useful overage — so a team that under-uses one product still pays for the discount it signed up for. Choosing PAYG ($8/GB, $1.5/1k) instead would have raised the marginal rate but removed the floor; the break-even is the whole game.

Archetype 2 — an AI agent product at scale. An agent company runs the Browser API for live page rendering and the SERP API for grounding, both on the Scale tier.

Here the lesson is bandwidth accounting: Browser API bills every byte the headless browser pulls (images, fonts, trackers), so the effective GB consumed is far higher than the HTML a scraper would fetch — making the Browser API the most expensive way to read a page unless full rendering is genuinely needed. This is the same “meter you don’t see until the invoice” risk covered in our guide to setting usage thresholds and spend alerts.

Want to estimate your own Bright Data bill? Use the Bright Data pricing calculator to model your monthly cost across proxy GB, per-IP allotments, and per-1k request volumes.

Pricing evolution : from Luminati proxies to AI-ready data

Bright Data’s pricing story is less about frequent rate moves and more about three structural shifts: the Hola→Luminati→Bright Data identity arc, the datasets model flipping from page-load quotes to a self-serve refresh dial, and the 2024–2026 reorientation around AI buyers. The hub pricing page (brightdata.com/pricing) is a product directory and carries no per-GB rates, so the dated price points below come from the per-product surfaces and Wayback snapshots of the datasets page; the proxy-network and scraper pricing pages were not preserved in the archive, so their historical rates are recorded as unknown rather than guessed.

Cadence

Tracked range: 2021-03 to 2026-06. “unknown” marks quarters where the price-bearing surface (proxy-network, scraper pages) was not preserved in the Wayback archive; only the datasets page and the products hub were captured, so per-GB and per-IP rate history could not be independently verified.

Notable changes

• 2015-05 — A 296-point Hacker News thread (2015-05-27) exposes that Hola VPN resells user bandwidth as exit nodes through its sister company Luminati — the origin of the residential network. Contemporary reporting put early residential bandwidth near $20/GB.
• 2017-08 — EMK Capital acquires Luminati out of Hola at roughly a $200M valuation, separating the B2B proxy business from the consumer VPN.
• 2021-03 — Luminati Networks rebrands to Bright Data, dropping the “Illuminati”-adjacent name and leaning into “web transparency” positioning.
• 2023-02 — Wayback snapshot of the datasets page shows “Starting from $0.001/record,” with price set by page-loads; today’s self-serve base is $2.50/1k records on a refresh-cadence dial — a 2.5× base increase and a model change.
• 2024-01 / 2024-05 — Bright Data wins the Meta scraping suit at summary judgment (Judge Chen, N.D. Cal.) and sees X Corp’s suit dismissed (Judge Alsup, N.D. Cal.) — rulings that legitimized the public-data-scraping model it sells (see below).
• 2025–2026 — Reported ARR crosses ~$300M (from ~$128M in 2024), powered by AI labs; product lineup reorganizes around agentic data access and the residential network scales from 72M to 400M+ IPs.

The scraping lawsuits in detail

Bright Data is unusual in the corpus for having a legal moat that doubles as marketing. In 2023 both Meta and X Corp sued it over scraping their platforms.

• Meta Platforms v. Bright Data — In January 2024, Judge Edward M. Chen (N.D. Cal.) granted Bright Data summary judgment on Meta’s breach-of-contract claims, holding that Meta’s terms bind only logged-in registered users and that scraping public, logged-out data fell outside them. Meta dropped the remainder of the case in February 2024.
• X Corp v. Bright Data — In May 2024, Judge William Alsup (N.D. Cal.) dismissed X’s claims, finding them preempted by the Copyright Act and warning that letting platforms over-control public data risks “information monopolies.”

Both outcomes reinforce Bright Data’s pricing thesis: if scraping public web data is legally defensible, then selling the infrastructure to do it at scale is a durable, meterable business — which is exactly why the company prices each collection surface on its own success-based unit.

What’s unique : a different meter for every product

Four simultaneous billing meters, one platform. Most usage-based vendors pick a single value metric and stretch it across the catalog. Bright Data does the opposite: rotating residential and the Browser API bill per GB, static ISP and datacenter proxies bill per dedicated IP, the unblocking/search/scraper APIs bill per 1,000 successful results, and datasets bill per 1,000 records. Each meter tracks the actual cost driver of that product, so margins stay legible even though the headline price list reads like four different companies. It is the cleanest multi-meter example in the corpus.

KYC compliance gating inside a self-serve funnel. Residential and mobile networks can’t be switched on until the account passes a know-your-customer review — sometimes including an intro video call. Embedding a manual compliance checkpoint in an otherwise instant, card-on-file usage product is rare, and it is a direct legacy of the Hola/Luminati origin and the “ethical sourcing” positioning the company adopted post-rebrand.

Refresh cadence as a discount axis. Datasets invert the usual “fresher = pricier” intuition: a One-time pull is the most expensive per record, and committing to Monthly refresh discounts the per-record rate by 80%. The cadence dial turns a subscription commitment into the lever that lowers unit price — a packaging idea more pricing teams could borrow, and one we explore in the usage-based pricing models overview.

“Pay only for success” on the request meters. Web Unlocker, SERP, and Web Scraper bill on successful results, not attempts — a block, CAPTCHA, or failed fetch isn’t charged. That outcome alignment is a real differentiator against proxy resellers that bill all traffic regardless of block rate, and it is the same logic behind the broader move toward outcome-based pricing in AI.

A deposit-match acquisition incentive. Rather than a flat free tier, Bright Data matches a first deposit dollar-for-dollar up to $500 in usable credits, plus a no-card 1,000-record trial on the Web Scraper API. It is a usage-credit-shaped acquisition lever — capital that only converts to revenue once the buyer actually runs traffic.

Strengths & weaknesses

Billing UX : spend limits, auto-recharge, and commitment top-ups

• Pay-as-you-go vs minimum monthly commitment — every product offers a no-commitment PAYG plan; committed tiers ($499 / $999 / $1,999) carry a minimum monthly account commitment billed on the 1st of each month.
• Usage spend limits per zone — under each proxy zone, a “Usage spend limit” column lets you cap daily usage by bandwidth (bytes) or money spent ($); the zone auto-suspends on reaching the cap (recalculated every 15 minutes).
• Auto-recharge — configurable balance top-up that triggers when available balance drops below 85% (or at 25% of a pre-set amount), to prevent suspension.
• Mid-month proration — joining mid-month prorates the first minimum commitment to the share of the month the account was active.
• Deposit-match promotion — first-deposit match, dollar-for-dollar up to $500, applied as usable credits.
• Free trial — Web Scraper API offers 1,000 records free for one week, no credit card; trial traffic is shown on the dashboard but not billed.
• Datasets refresh dial — a refresh-rate selector (One-time / Biannual / Quarterly / Monthly) and a records slider drive a live total-price calculator.
• Payment methods — PayPal, Payoneer, Alipay, Google Pay, wire transfer, credit card, plus AWS Marketplace and Azure billing.

Strategic wins : matching meter to value

1. A purpose-built meter for each product

Bright Data bills bandwidth-heavy rotating proxies per GB, static IPs per dedicated IP, unblocking/search APIs per 1,000 successful results, and datasets per 1,000 records. Each meter tracks the underlying cost driver of that product, which keeps margins predictable and prices defensible — a clean illustration of the value-metric problem solved one product at a time rather than forced onto a single unit.

2. A PAYG on-ramp under every product

Every product exposes a no-commitment pay-as-you-go rate before any committed tier, so a developer can start at $8/GB or $1.5/1k requests without sales contact. This lowers acquisition friction and lets usage prove value before a buyer commits to a $499+ monthly floor.

3. “Pay only for success” framing on request-based APIs

Web Unlocker, SERP, and Web Scraper bill only on successful results/records, not attempts. Charging for outcomes rather than effort de-risks the buy and differentiates against proxy resellers that bill all traffic regardless of block rate.

Areas to improve : cross-product cost legibility

1. Make blended cross-product cost estimable

With four meters in play, a buyer combining residential GB, Web Unlocker requests, and a datasets subscription cannot easily forecast a single monthly bill. A unified, public multi-product cost estimator (or a clearer “what a typical stack costs” worked example) would reduce pre-purchase uncertainty.

2. Publish mobile-proxy pricing or label the gate clearly

Mobile proxies are referenced in the FAQ but carry no public per-unit price on the proxy-network page, forcing a contact-sales detour. Either a public starting rate or an explicit “KYC-gated, contact us” card would set expectations instead of leaving the meter invisible.

3. Surface minimum-commitment floors earlier

The committed tiers ($499 / $999 / $1,999) are minimum monthly commitments, not just discounts, with mid-month proration and top-up mechanics buried in the FAQ. Foregrounding the floor cost on each pricing card would prevent surprise at the first invoice.

Key takeaways

1. Match the meter to the cost driver, even if it means many meters. Bright Data runs four value metrics at once (per-GB, per-IP, per-1k-result, per-1k-record) because each tracks a genuinely different cost. The clarity-vs-simplicity tradeoff is real, but a meter that mirrors cost keeps margins defensible — see our guide to choosing the right usage metric.
2. A committed tier is a floor, not a discount — say so. The $499/$999/$1,999 plans are minimum monthly commitments with proration and top-up rules. Teams that surface “lower rate” without surfacing “minimum spend” set buyers up for invoice surprise; foreground the floor on the card.
3. Account for the bytes the buyer can’t see. Bandwidth is billed on full request+response, and the Browser API bills every asset a headless browser loads. Whenever the meter counts more than the “useful” output, the pricing page should teach the counting rule, or the first invoice will.
4. Outcome-based billing is a competitive wedge, not just fairness. “Pay only for successful results” both de-risks the buy and structurally beats resellers who bill failed traffic. Charging for outcomes is a positioning move as much as a packaging one.
5. Legal defensibility can be a pricing asset. By winning the Meta and X scraping suits, Bright Data turned a regulatory risk into a moat — its prices are credible precisely because the activity they meter has been tested in court. Trust and compliance posture (why finance teams care) are part of the price.

UBP implications

1. Per-product value metrics beat a forced-uniform meter for multi-product platforms. Bright Data shows you can run several meters coherently if each maps to a real cost — the cost to buyers is forecasting complexity, which a calculator or worked examples can absorb. The lesson generalizes to any vendor with structurally different product lines.
2. Commitment floors are the quiet bridge from pure-usage to hybrid. Layering minimum monthly commitments under metered overage converts “pure pay-as-you-go” into a hybrid model with predictable revenue — useful for vendors who need both PLG on-ramps and forecastable ARR.
3. As AI agents become the buyer, success-metered access becomes the unit of value. Bright Data’s pivot to agentic surfaces (Search & Extract, Agent Browser, per-1k results) shows the meter that matters for AI agents is successful data acquisitions, not seats or flat subscriptions — a signal for anyone pricing infrastructure that agents consume.

Sources

• Bright Data — Products pricing overview (accessed 2026-06-04)
• Bright Data — Datasets pricing (accessed 2026-06-04)
• Bright Data — Bright Insights pricing (accessed 2026-06-04)
• Bright Data — Proxy Networks pricing (residential / ISP / datacenter tabs) (accessed 2026-06-04)
• Bright Data — Browser API / Scraping Browser pricing (accessed 2026-06-04)
• Bright Data — SERP API pricing (accessed 2026-06-04)
• Bright Data — Web Scraper API pricing (accessed 2026-06-04)
• Bright Data — Scraping Functions (IDE) pricing (accessed 2026-06-04)
• Bright Data — Web Unlocker API pricing (accessed 2026-06-04)
• Bright Data — Why we changed our name from Luminati Networks to Bright Data (accessed 2026-06-04)

Bottom line

Bright Data runs one of the clearest examples of per-product value-metric selection in the corpus: bandwidth for rotating proxies, dedicated IPs for static ones, successful requests for unblocking, and records for datasets — each metered where the cost actually lives. The trade-off is that no single number describes the platform, and budgeting means tracking four meters at once. See the rest of the Blueprint corpus for how peers handle multi-product usage pricing.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Current per-product pricing snapshot

Jun 2026

Proxies per-GB (residential $8→$2.50/GB) and per-IP (ISP $1.8→$1.3, datacenter $1.40→$0.90); Web Unlocker / SERP / Scraping Functions per-1k results at $1.5→$1.00; Browser API $8→$5/GB; Web Scraper API $1.5/1k records PAYG with a $499/mo Scale tier; Datasets from $2.50/1k records; Bright Insights from $250–$400/mo. A 50% residential promo and 25% Scraper API promo were live at capture.

AI-data pivot: 400M+ IPs, agentic surfaces, 50% residential promo

May 2026

By the 2026-05-01 snapshot the hub advertises 400M+ residential IPs (up from 72M in 2024) and a reorganized lineup oriented around AI: Discover API, Browser API, Scraper Studio, a +50PB Web Archive, Deep Lookup, Agent Browser, Search & Extract, and Data-for-AI packages (incl. video feeds for VLA/robotics). A '50% OFF' residential promo runs alongside.

Products-hub pricing page; 72M residential IPs

Jun 2024

The brightdata.com/pricing hub (2024-06-07 snapshot) is a product directory linking out to per-product pricing — no per-GB rates on the hub itself. Residential network advertised at 72M+ IPs; '20,000+ customers worldwide.'

Datasets priced at $0.001/record, page-load driven

Feb 2023

A Wayback snapshot (2023-02-06) shows the Datasets page 'Starting from $0.001/record' ($1.00 per 1,000 records), with price 'determined by the number of page loads required to collect, build, and maintain the dataset.' Custom Datasets (managed) started at $1,000/mo annual or $1,500 one-time. Scale shown then: 72M residential, 7M mobile, ~770K datacenter IPs.

Luminati Networks rebrands to Bright Data

Mar 2021

The company dropped the 'Luminati' name (which invited unwanted 'Illuminati' associations) and rebranded to Bright Data, leaning into 'web transparency' positioning. The product/meter structure was unchanged; the change was identity and compliance-led.

EMK Capital acquires Luminati (~$200M)

Aug 2017

London PE fund EMK Capital bought Luminati Networks out of Hola at a ~$200M valuation, separating the B2B proxy business from the consumer VPN. This set up the eventual rebrand and the compliance/KYC posture.

Hola/Luminati era — residential bandwidth resold from a VPN userbase

May 2015

The residential proxy network launched as Luminati, monetizing the idle bandwidth of Hola VPN's free users as exit nodes. A 296-point Hacker News thread (2015-05-27) exposed the model; contemporary reporting put early residential bandwidth around $20/GB. (Secondary-sourced; original Luminati pricing pages were not captured in this pass.)