Rayobyte Review
Genuinely fast and accurate in the United States, on a pool that thins sharply everywhere else.
The short version
- Rayobyte answered faster than us on median in four of five markets and had the tighter p95 in all five.
- It served 119,144 live IPs across five countries, with 86,099 of those in the United States alone.
- Outside the US the pool is thin: 4,429 live addresses in Germany and 5,438 in France.
- Geo accuracy is excellent in four markets, 99.6% to 99.9%, but drops to 94.9% in Germany.
- It completed 92.5% to 94.0% of the standard test, with redirects the largest single cause.
At a glance
What we measured
Live IPs each network had answering during the same test window, five countries, identical request volumes and concurrency for both sides.
| Country | Rayobyte | Shifter | Networks |
|---|---|---|---|
| United States | 86,099 | 136,667 | 1,266 |
| United Kingdom | 14,345 | 53,506 | 193 |
| France | 5,438 | 33,581 | 123 |
| Canada | 8,833 | 33,485 | 183 |
| Germany | 4,429 | 22,005 | 272 |
It is faster than we are
This is the headline result, and it is the first time in this series we have had to write it. Rayobyte answered in 333 ms in the United States against our 364 ms, and led on median in Britain, France and Germany too. Only Canada went our way, 324 ms against 375 ms.
The consistency is more striking than the median. Its p95 was better in all five markets, and by a wide margin in two of them: 1,191 ms against our 4,720 ms in Britain, and 1,415 ms against 4,271 ms in Germany. A tight p95 is what stops a crawl stalling on stragglers, and Rayobyte has one.
If your workload is latency sensitive and United States based, this is a genuinely good network and we would not argue with anyone choosing it on that basis.
The pool thins sharply outside the United States
The US result is respectable: 86,099 live addresses across 1,266 operators, about 63% of what we returned in the same window. That is a real pool.
Everywhere else it drops away. Germany returned 4,429 live addresses against our 22,005, and France 5,438 against 33,581. Britain and Canada sit around a quarter of our counts. For a network advertising 40M+ IPs across 163+ locations, the measured depth outside its home market is modest.
Network spread tells a different and more flattering story. Britain was a dead heat at 193 operators each, and Germany was near level at 272 against our 276. So the addresses Rayobyte does hold are spread across a comparable range of ISPs; there are simply fewer of them live at any moment.
Geo accuracy, and a figure we are not quoting
Four of five markets came back between 99.6% and 99.9%, and the United States at 99.9% was better than our own 97.4%. Germany is the exception at 94.9%, meaning one request in twenty exited outside the country requested.
We ran Rayobyte once before, on 2026-08-13, through a different gateway of theirs, and measured 74.7% US geo accuracy. On the endpoint tested here the same measurement is 99.9%. We treat the earlier number as an artifact of that endpoint rather than a property of the network, and we do not quote it anywhere on the site.
The lesson is worth stating plainly for anyone reproducing our work: on Rayobyte the gateway you choose changes the result materially, so name the endpoint when you publish a number.
Where it struggles
It does not complete the test. Across the five markets it answered between 92.5% and 94.0% of requests. The largest single cause was redirects, between 1,957 and 7,112 per run, with timeouts and 503s making up most of the rest.
A separate sticky-session test at 1,000 concurrent produced 647 outright 403 responses on first contact, which reads more like an account concurrency ceiling than a network fault. We report it because it is what we saw, not as a settled diagnosis.
Single requests succeed reliably and quickly. The shortfall appears under load, which is exactly where a scraping workload lives.
Pricing
Rayobyte starts at $3.50 per GB and steps down through $2.00 at 50 GB, $1.50 at 250 GB and $0.70 at 1,000 GB, reaching $0.50 only at 5,000 GB, a $2,500 purchase.
At 100 GB, a volume a working team actually buys, it is $2.00 against our $1.00. Our best rate is $0.75 from 200 GB, which most buyers reach; theirs is $0.50 but requires fifty times that commitment.
So the honest summary on price is that Rayobyte has the lower floor and the higher real-world cost. If you genuinely buy 5 TB a month, their rate beats ours and we would tell you so.
Pricing
| Plan | Traffic | Per GB | Monthly |
|---|---|---|---|
| Starter | 1 to 49 GB | $3.50 | $3.50 to $171.50 |
| Consumer | 50 to 249 GB | $2.00 | $100 to $498 |
| Professional | 250 to 999 GB | $1.50 | $375 to $1,498.50 |
| Corporate | 1,000 to 4,999 GB | $0.70 | $700 to $3,499.30 |
| Enterprise | 5,000 GB and up | $0.50 | $2,500 and up |
Pros and cons
- Faster than Shifter on median in four of five markets
- Tighter p95 than Shifter in all five, notably 1,191 ms against 4,720 ms in Britain
- 99.9% US geo accuracy, better than our own 97.4%
- Matched our British network spread exactly at 193 operators
- Country, state, city and ASN targeting at no extra cost
- Lowest published floor rate in this series at $0.50/GB
- Pool thins sharply outside the US: 4,429 live IPs in Germany, 5,438 in France
- Completed only 92.5% to 94.0% of the standard test, mostly redirects
- German geo accuracy of 94.9%, well behind its other markets
- $2.00 per GB at 100 GB against $1.00, and the $0.50 floor needs $2,500 up front
- Targeting parameters go in the password, which trips up standard proxy clients
Who Rayobyte is for
- Your work is United States focused, where its pool is deep and its geo accuracy beat ours.
- Latency matters more to you than pool size, since it is one of the quicker networks we have measured.
- You want predictable response times, because its p95 was tighter than ours in every market.
- You need depth outside the US, where it returned three to six times fewer addresses than we did.
- You run at high concurrency, where between 6% and 7.5% of requests did not complete.
- You buy at ordinary volumes, at $2.00 per GB against $1.00 for the same 100 GB.
How we scored it
| Criterion | Weight | Score | Basis |
|---|---|---|---|
| Live pool measured | 25% | 3.5 | 119,144 IPs live across five countries, against the tested field. |
| Network spread measured | 20% | 7.9 | 2,037 distinct networks across five countries. |
| Price per GB measured | 25% | 7.1 | $2.00 per GB at 100 GB, the volume a working team actually buys. The $0.50 floor requires a 5,000 GB purchase at $2,500, and the entry rate is $3.50. |
| Response time measured | 10% | 10 | 381 ms median across the countries tested. |
| Country coverage | 10% | 8.4 | 163+ countries advertised, scaled against full global coverage. |
| Features | 5% | 8 | HTTP and HTTPS, rotating or sticky sessions, and country, state, city and ASN targeting included at no extra cost. ASN and state control put it ahead of most of this series. No SOCKS5 on the residential product. |
| Transparency | 5% | 8 | The full price ladder is published without a sales gate, and per-country pool claims are broken out rather than given as one headline number. Against that, the advertised 40M+ is a long way from what any single window measures. |