Thordata Pricing Breakdown: How Much Should You Really Pay for Scraping Proxies?

Proxy pricing is one of those categories where the sticker price tells you almost nothing about the real bill. Two teams can use the same provider, scrape the same kind of sites, and walk away with invoices that differ by 5x. That is not a scam, that is just how per-GB proxy billing interacts with modern JavaScript-heavy pages.

This post is a hands-on breakdown of what Thordata actually costs in practice, how to model your own bill before you swipe a card, and when a cheaper datacenter pool is a trap vs. when residential is worth the premium.

How proxy pricing actually works (and why the per-GB model wins)

Most serious proxy networks, Thordata included, charge per gigabyte of bandwidth, not per request and not per IP. That sounds straightforward, but it has three big consequences you need to internalize before you look at a pricing page.

• Your bill scales with page weight, not page count. A 50 KB HTML page and a 3 MB JS-rendered product listing both count as "one request" in your code, but one costs 60x more on your proxy invoice.
• Failed requests still cost money. If a target returns a 403, a CAPTCHA page, or a 10 MB anti-bot challenge, that bandwidth is on your tab.
• Retries multiply the bill. A scraper that retries 3 times on failure can triple your bandwidth consumption on fragile targets.

This is why a "cheap" per-GB price can end up more expensive than a higher one. What you care about is total landed cost per successful record, not the headline rate.

Per-GB vs per-IP vs per-request: which model is Thordata using?

Thordata uses per-GB for residential, mobile, and ISP pools, and offers per-IP pricing on static ISP and dedicated datacenter plans. This is the industry norm. The per-IP model is usually a better fit for use cases where you want a stable identity (account management, ad verification, sneaker copping). Per-GB is the right fit for large scraping jobs where you rotate heavily and do not care which IP you land on.

If you are unsure which pool fits, our web scraping and proxy category walks through the tradeoffs with side-by-side tool comparisons.

What Thordata actually charges (benchmark ranges)

Thordata publishes tiered pricing that drops sharply as your committed volume goes up. At the time of writing the ballpark looks like this, and we will update as plans change.

• Residential proxies: roughly $2.50 to $4 per GB at entry tiers, dropping toward $1 per GB at high volume
• Mobile proxies: typically $6 to $8 per GB, with steep discounts past 100 GB
• ISP (static residential) proxies: per-IP pricing, usually a few dollars per IP per month for monthly rentals
• Datacenter proxies: the cheapest option, usually well under $1 per GB and/or cheap per-IP pools

Compared to the incumbents, this puts Thordata in the aggressive-challenger bracket. Bright Data and Oxylabs tend to sit at the top of the market, SmartProxy/Decodo sits in the middle, and Thordata undercuts both on residential while still offering the full pool variety you need for serious scraping. If you want a more holistic comparison, see our roundup of the best residential proxies for web scraping.

Free trials and pay-as-you-go

Thordata offers free trials on most pool types, which is genuinely useful because you can actually measure your bandwidth-per-record before you commit to a plan. Do not skip this step. A 500 MB trial on your real targets is worth more than any spec sheet.

A realistic monthly bill: 1M, 10M, and 100M requests

This is where most pricing articles wave their hands. Let us do actual math. We will assume a mixed workload scraping e-commerce product pages, where each request pulls an average of 600 KB of response bandwidth (including some JS-rendered pages but not the heaviest).

Scenario A: 1 million requests per month

• Bandwidth: 1M requests x 600 KB = roughly 600 GB
• Wait, that is already a huge number. Right. One million JS-heavy pages is not a small job.
• At $2.50/GB residential: about $1,500/month
• At $0.80/GB datacenter (for sites that do not need residential): about $480/month

If your target site will accept datacenter traffic, you just saved 70%. That decision is the single biggest lever you have.

Scenario B: 10 million requests per month

At this tier you will be on a volume contract, so assume residential drops to about $1.80/GB.

• Bandwidth: 10M x 600 KB = 6,000 GB = 6 TB
• Residential at $1.80/GB: about $10,800/month
• Datacenter at $0.50/GB: about $3,000/month

You will also want to think hard about your retry budget. A 5% failure-with-retry rate adds roughly 5% to the bill, which at this scale is $500+. Fixing your scraper's error handling is now a legitimate cost-savings project.

Scenario C: 100 million requests per month

At this volume, you are negotiating. Published rates do not apply. You will be talking to a rep about a commitment-based contract, likely with a landed rate closer to $1/GB on residential and $0.20 to $0.30/GB on datacenter.

• Residential at $1/GB: about $60,000/month
• Datacenter at $0.25/GB: about $15,000/month

At this scale, optimizing your bandwidth footprint (compression, blocking images/fonts, using the HTML-only endpoint where the target allows it) can save you more than switching providers.

The hidden costs nobody talks about on the pricing page

The posted per-GB rate is the floor, not the ceiling. Here are the line items that inflate real-world bills.

1. JS-rendered pages cost 10-50x more than HTML

A static HTML product page is 30-80 KB. The same page rendered with JS, including all the analytics scripts, video players, and third-party embeds, can be 2-5 MB. If you are scraping via a headless browser and not blocking non-essential resources, your bandwidth bill is mostly ads and telemetry, not data you care about.

Fix: Always block images, fonts, media, and third-party scripts in your headless config unless you specifically need them. This alone can cut your bill by 60-80%.

2. Failed requests and CAPTCHA pages

Anti-bot challenges are not free. A Cloudflare interstitial is a full page load. If 20% of your requests hit a challenge, you are paying for 20% wasted bandwidth that returned zero data.

Fix: Use a session-aware scraper that backs off when it starts seeing challenges instead of hammering through them. And pick the right pool, residential fails less often on protected targets.

3. Retry loops

A naive retry-3-times-on-any-error policy turns a 10% failure rate into a 30%-ish bandwidth overhead. At scale this is real money.

Fix: Retry only on transient errors (timeouts, 5xx). Do not retry on 403 or 429 without rotating identity and backing off first.

4. Overage rates

Most providers, Thordata included, charge a premium when you exceed your committed tier. Overage rates can be 1.5-2x the in-plan rate. If your traffic is bursty, size your plan for the peak, not the average.

When residential is worth the premium over datacenter

Residential IPs can cost 3-5x more per GB than datacenter. That premium is worth it only when the target actively fingerprints datacenter ranges. The honest rule of thumb:

• Datacenter works fine for: public APIs, most SEO tools, smaller e-commerce sites, archival sites, anything without aggressive anti-bot
• You need residential for: major retailers (Amazon, Walmart, Target), travel and airline sites, social platforms, sneaker/ticket sites, ad verification on major networks
• You need mobile for: consumer mobile-only apps, Instagram/TikTok-grade anti-bot, carrier-detected targets

If you are not sure, run a 1,000-request test on your target with each pool and measure the success rate. The math usually makes the choice obvious. For more on how to pick, see our guide to the best web scraping tools and our comparison of proxy types.

How to model ROI before you commit

The question is never "is Thordata cheap?" It is "does the data I pull pay for the bandwidth I burn?" A quick framework:

1. Cost per successful record. Total proxy spend divided by records successfully extracted and stored. This is your true KPI.
2. Revenue (or value) per record. If you are reselling data, this is the price. If you are using it internally, estimate the business value.
3. Margin. Value per record minus cost per record. If this is negative or razor-thin, either the target is wrong or your scraper is inefficient.

Do this math at a small scale (a few thousand records) before you spin up a 100-million-request job. Most teams that blow up their proxy budget skipped this step.

Frequently Asked Questions

Is Thordata cheaper than Bright Data and Oxylabs?

On residential, yes, Thordata's headline rates undercut both meaningfully at entry and mid tiers. At very high volume the gap closes because the enterprise players will negotiate aggressively. On datacenter the difference is smaller because all providers are cheap there.

What is a realistic per-GB rate for residential proxies in 2026?

$2 to $4 per GB is normal at entry tiers, $1 to $1.50 per GB at volume, and sub-$1 only on very large commitments. If someone quotes you $0.50/GB residential with no commitment, be skeptical about pool quality.

How do I estimate my bandwidth usage before buying?

Run a 500-1,000 request pilot on your actual target pages using a free trial. Record the total bytes transferred, divide by the number of requests, and project. Then add 20% for retries and failures.

Can I use Thordata datacenter proxies for Amazon or Google?

You can try, and on Google search it often works. On Amazon you will hit walls quickly. Use residential for any major retailer with serious anti-bot infrastructure.

Does Thordata charge for failed requests?

Yes. Like essentially every per-GB proxy provider, you pay for bandwidth consumed, not for successful extractions. This is why scraper quality directly impacts your bill.

How do I reduce my proxy bill without dropping volume?

In this order: block non-essential resources in your headless browser, reduce retries, switch from residential to datacenter on any target that accepts it, cache responses where possible, and upgrade your tier to hit a lower committed rate.

Is per-GB or per-IP pricing better for scraping?

Per-GB for rotating, high-volume scraping. Per-IP for stable-session work like account management, ad verification, or any task where you need a specific IP to stay alive for hours or days.

The bottom line

Thordata's pricing is competitive, but the real variable in your bill is not the provider's rate card, it is your scraper. A well-engineered pipeline on a mid-priced provider will beat a sloppy pipeline on the cheapest provider every time.

If you are scoping a new project, start with the free trial, measure your actual bandwidth per record, pick the cheapest pool that clears your target's defenses, and then negotiate volume. That is the playbook. If you want to see how Thordata stacks up against alternatives in practice, check our residential proxy roundup and the full web scraping and proxy category.