Residential proxy prices stopped falling in 2026. The cheap entry tier went with them.

For three years, the residential proxy price chart pointed one way. Between 2023 and 2025 the headline rate on a gigabyte of residential traffic fell by as much as 75 percent, according to Proxyway's 2026 market research. Teams collecting web data at scale planned around that. Sign the smaller contract now, the thinking went, because the same bytes will be cheaper at renewal.

In 2026 the line flattened and, at the top of the market, turned back up. The same research found residential rates had stabilized and started to revert. It did not arrive as a price increase email. It arrived as discount codes that stopped renewing and entry plans that quietly left the pricing page.

Did residential proxy prices go up in 2026?

For buyers at the small end, effectively yes, even though the sticker rate mostly held. The major networks removed the standing 40 to 50 percent coupon codes that had made their list prices academic, and several dropped the pay-as-you-go plans that let a team start without a commitment. Proxyway reports the coupon pulls at IPRoyal, Decodo, Oxylabs and Massive, and pay-as-you-go plans discontinued at Oxylabs, Massive and SOAX. If you were paying a post-coupon rate last year, your bill went up while the number on the page did not.

The coupons came off

The residential price war was fought with coupons more than with list prices. A vendor could keep a respectable rate on its pricing page, then hand out a 40 or 50 percent code that set the real number well below it. That let everyone advertise stability while competing hard underneath. In 2026 the codes came off.

The detail is worth keeping straight, because it cuts against a simple "prices went up" headline. Proxyway reports that Decodo and Oxylabs actually cut their permanent list plans by around 25 percent, so on paper those rates fell. With the long-running coupons gone, though, the price a returning customer pays still rose. IPRoyal skipped the halfway step and reverted to its pre-coupon rates outright. Different mechanics, same result: the effective rate went up even where the list rate went down.

A worked example makes the shape clear. Say a network listed residential at $4 per gigabyte through 2025 and ran a standing 50 percent code, so the rate you actually paid was $2. In 2026 the vendor drops the list price to $3, a real 25 percent cut on paper, and retires the code. Your new rate is $3, up 50 percent on what you paid last year, announced as a price reduction. Nobody has to be dishonest for that to happen. The list price and the paid price were never the same number, and only one of them was ever on the page.

This is the recurring trap with per-GB pricing. The unit stays fixed and legible while the discount layer under it, which is what actually sets your invoice, moves without notice. It is the same problem covered in per-GB pricing stopped describing what you pay for, seen from the other side: last year the hidden movement was downward, this year it reversed.

The entry tier moved out of reach

The change that lands hardest on smaller teams is not the rate. It is the entry point. Oxylabs, Massive and SOAX discontinued pay-as-you-go residential plans, and Proxyway describes the effect as raising the entry point by up to twenty-five times. A buyer who could previously put down a few dollars for a couple of gigabytes, to trial a pool or cover a one-off job, now meets a minimum commitment at the door.

For anyone still choosing a vendor, that matters more than a few cents per gigabyte. Evaluation gets more expensive on both sides of a comparison. Running a split test between two residential proxy pools on your own targets is the only measurement that tells you what a pool costs per usable record, and it now carries a real upfront cost per vendor rather than a trivial one. When the cheap way to compare disappears, teams default to the incumbent, which is usually the outcome the incumbent's pricing change was going to produce anyway.

The floor kept falling, just not for everyone

Calling 2026 a clean price increase would be wrong. Underneath the premium tier, some vendors kept cutting. Proxyway reports ProxyEmpire and Webshare roughly halving their rates against March 2025, and grey-market operators pricing the whole field down to 15 to 50 cents per gigabyte. Those bottom rates are a warning rather than a deal. Bandwidth that cheap usually means a pool whose IP sourcing will not survive a look, which is a procurement question about where the addresses come from before it is a price question.

Mobile went the other way entirely. Rates there were cut by as much as 98 percent, a move Rayobyte led earlier in the year. Bright Data closed its mobile proxy product to new customers in April 2026, phasing the line out instead of competing on it. So residential firmed up at the premium end while the categories on either side of it, budget residential and mobile, were still falling or being abandoned. The market did not simply get more expensive. It split.

Why the decline reversed

A few things line up. The coupon war was a loss leader, and loss leaders end. The consolidation that gave nearly every proxy network a parent company left fewer independent vendors with a reason to keep undercutting each other. And residential bandwidth has a real cost floor: sourcing consented IPs, running the gateway, and keeping a pool clean all cost money the rate cannot drop below without the product getting worse. Most vendors in Proxyway's survey expected prices to stabilize for that reason. The cheap end had reached the point where another cut meant a worse pool.

None of this makes residential proxies expensive. It makes the multi-year discount trend something you can no longer extend into next year's budget by drawing a straight line.

What it changes at renewal

A few things follow for a team buying collection capacity.

Stop pricing the next contract on continued decline. The assumption that residential gets cheaper every year held for three years and does not hold now. If your usage is growing and you can forecast it, a committed plan locked at today's rate is easier to defend than waiting for a cut the market has stopped expecting. Bright Data's own residential pricing runs about $8.40 per gigabyte pay-as-you-go down to roughly $3 on commitment. That spread is the lever, and it now favors committing sooner than it did a year ago.

If you leaned on pay-as-you-go to trial pools or absorb traffic spikes, that door is closing at the majors. Budget for a minimum commitment when you plan a vendor trial, or run the trial at a vendor that still sells in small amounts. The guide on evaluating web data providers covers how to structure a trial that ends in a decision rather than an open tab.

Price the tiers separately instead of shopping one number. Premium residential that is firming up and a budget pool at a fifth of the price are not substitutes. They fail on different targets. Match the pool to the defense on the target, along the lines in picking a proxy type by target, and read the grey-market rate as the flag it is. The billing models themselves sit in proxy pricing explained.

Keep a second vendor live, even a small one. A rising entry point cuts both ways: it makes casual switching costlier, which is the moment a standing relationship with a backup pool earns its minimum. A second vendor already carrying real traffic gives you a price reference the next renewal conversation cannot wave away, and somewhere to move if a pool degrades after an acquisition. It is the same dual-vendor arrangement described in running two proxy vendors, and in a market that has stopped competing on coupons, the leverage it buys is worth more than it was when a discount code could be had for asking.

The one-directional residential price chart was a useful planning tool while it lasted. In 2026 it stopped predicting anything. The rate is no longer the story. Where the entry point sits, and which tier your targets actually need, is.