An unlimited bandwidth residential proxy is unlimited in gigabytes and limited somewhere else. A residential seller pays for every gigabyte that crosses somebody's home connection, so a flat price only works if the plan caps something the seller can control: how many connections you keep open, how many ports or addresses you get, how fast each one goes, or a fair-use clause that lets the seller decide later. The plan is cheaper than paying per GB only if that other limit is one your work never reaches.
Where the limit goes
Read the terms of an "unlimited" plan for four things. Expect to find at least one of them:
- a cap on threads, meaning connections open at the same time, so traffic is free and parallelism is not
- a fixed number of gateway ports or static addresses, each one a single exit, with more exits costing more
- a line rate on each port or on the whole plan, so heavy use gets slow rather than refused
- a fair-use clause with no stated cap, under which the seller can throttle or suspend usage it considers excessive
The last one is the hardest to plan around. When the clause has no number in it, the limit is real and you do not know where it is.
A fifth thing to check is what kind of proxy it is. Some plans sold as unlimited residential turn out to be static ISP addresses: residential-looking IPs hosted in a data centre, sold per address. That is a different product with different strengths, and it is priced like one (how per-GB and per-IP pricing compare has the arithmetic).
Paste one gateway line. Every profile mints its own exitIn Mosaic: Provider minting. One gateway line, a separate exit for every profile. Free holds 3 profiles. Pro is $49 a month for any number.
Mosaic is a desktop app
Email yourself the link and open it on your computer.
How it works: Provider minting
A thread is not a browser
The cap that surprises people running browsers is threads, because a thread is usually counted as one open connection, and a browser is not one connection.
Chrome opens up to six connections to each host for HTTP/1.1, and an ordinary page pulls from dozens of hosts: the site itself, its image servers, fonts, analytics, a tag manager. HTTP/2 and HTTP/3 fold many requests into one connection per host, which helps, but a single page load still holds a connection per host it touches at its busiest moment, and a browser keeps some of them alive for reuse afterwards. A plan with a hundred threads is a few browsers at a time, not a hundred.
So the question to ask about a thread-capped plan is how many profiles you run at once, at their busiest, not how many you own. In Mosaic every profile is a separate browser with its own proxy connection, so ten profiles open side by side are ten browsers' worth of connections through the proxy, whatever each one is doing. Profiles you have closed use nothing.
Bandwidth is the cheaper number to know
Gigabytes are easier to measure than threads, because a browser's traffic is mostly the pages it loads, and the seller's own usage page counts them. An order dashboard is light; a feed that autoplays video is many times heavier. Run one real profile on its usual work for a week, read the seller's meter at both ends, and multiply by the profiles you open (how much proxy traffic a profile uses has the method with DevTools as a cross-check). That gives you a monthly figure to set against a per-GB price.
If that figure is small, a pay-per-GB plan is likely cheaper than any flat plan, and it has no thread cap to plan around. Unlimited plans earn their price at steady, heavy volume, and on work that can live within the plan's concurrency.
What to read before paying
- Find the word the plan uses for its cap: threads, connections, ports, IPs or "fair use".
- If it is fair use, look for a number. No number means the seller decides.
- Check whether "residential" means rotating home connections or static ISP addresses.
- Check whether sticky sessions are offered and how long one lasts, because a profile needs the same exit for its whole session.
- Run the proxy before you trust an account to it.
That last step is the one we do for every proxy, flat-priced or not. Mosaic checks a proxy before every launch and refuses to open a profile behind one that has not answered and tunnelled, and the check records the latency it measured. A throttled port shows up there as a slow path rather than as a mystery later (how to check a proxy before you use it covers the rest of the list). Throttling that only arrives after heavy use is the one limit a single check cannot see, which is why the fair-use clause is worth reading before you buy.
