Byteful: the fastest nodes in the market, and the hardest walls

Four years of scraping has taught me to distrust proxy math that looks too clean. SERPwatch ran 2M SERP checks a month on a startup budget, and the lesson stuck: a proxy that tests clean can collapse the moment customers onboard. So when I put six providers through the same 20-site probe set in September 2026, I was hunting the one that would actually get me data on the sites I care about. Byteful came back with the fastest nodes I measured (961ms median latency, 1.61 RPS, a US gateway at 750ms) and simultaneously the lowest success rate: 69.3%, with 0% on zillow and 5% on ebay. The fastest pipe in the market, and the fastest way to get blocked on the sites that matter. Speed without success is noise.

Before I ever reached the network, the product nearly stopped me. Google SSO signed me up but left me unable to sign in: dashboard.byteful.com routed to the sign-up page, so I reset my password, waited for a 2FA email, and logged in the hard way. The 1GB free trial is the most generous cap of the six providers, but it's locked behind Idenfy KYC that needs a camera. My Mac mini couldn't run it, and when I switched to a MacBook with a camera, Idenfy hit me with "Token does not exist." I gave up on the trial, bought the 2GB/$6.50 package, and found the $3.25 entry price is unreachable without loading $5 into a balance first. Every step had a lock.

• 961ms p50: the fastest nodes of all 6 providers, ~220ms faster than oxylabs (1,180ms)
• 1.61 RPS: the highest throughput of all 6, ahead of decodo (1.55)
• 69.3% success: the lowest of all 6; block rate 29.3% is the highest too
• Success climbs under load: 62.5% at c1 to 73.5% at c8, the biggest c1-to-c8 climb of the set, with p50 flat at ~900ms
• Hard walls: zillow 0%, tripadvisor 0%, yelp 0%, ebay 5% (worst of all 6)
• The catch: the onboarding hides everything: pricing behind buttons, a $5 minimum balance, KYC for the trial

The only provider that gets faster under load

Most providers plateau or degrade as concurrency rises. Byteful did the opposite: success climbed from 62.5% at c1 to 63.1% at c4 to 73.5% at c8, the biggest c1-to-c8 gain of all six providers. The latency stayed flat (928ms to 906ms to 924ms) and throughput scaled cleanly: RPS went 0.60 to 2.07 to 2.33 as I pushed it. The nodes don't just start fast; they hold speed while everything else on the network is competing with you.

On easy sites, that speed shows. agoda 100%, amazon 99.2%, imdb 99.2%, ozbargain 99.2%, reddit 98.5%, booking 97.7%. For a high-volume crawl of sites that don't fight back, this is the fastest transport I measured, 2.5x faster than shifter's 2,402ms and dataimpulse's 2,114ms median.

The limits of the speed show up the same way. macy drops to 32% on Byteful's fast tier where other providers hold 61-72%. expedia is weak for everyone (8-30%): anti-bot defeats rotation broadly. Fast nodes only help when the site lets you through.

The walls that 961ms can't touch

The same pool that clears agoda at 100% gets 0% on zillow, tripadvisor, and yelp, and 5% on ebay, the worst ebay result of all six providers. These aren't speed problems. The auto-tier browser stack (camoufox then playwright) rescued other providers on the same sites: on zillow, decodo got 73%, dataimpulse 95%, shifter 96%, all with the same transport I gave Byteful. Byteful stayed at 0%. That's not my setup. That's exit-node reputation.

The one bright spot on the tough list: slickdeals at 26%, the best result of all six providers (dataimpulse, the next best, got 19%; shifter 11%; decodo, oxylabs, webshare 0%). Reputation is per-site, not global. The pool that is dead last on ebay is best-in-class on slickdeals.

The lesson I keep landing on, again: pool size is marketing, exit-node quality is what matters, and Byteful's exit nodes carry the fastest pipe and the weakest reputation on the sites that actually gate you.

The 15-minute login loop

The sign-up flow only has a "sign up" button: there is no sign-in button on the page you land on. dashboard.byteful.com routes to sign-up again instead of sign-in. With Google SSO it was worse: instead of a natural re-auth, I had to reset my password openly, get a 2FA code by email, and log in that way. Fifteen minutes of rerouting for a one-minute task.

The 1GB free trial is genuinely the most generous of the six providers (decodo gives 300MB with a 3-day cap, webshare gives flunky free datacenter IPs). But Byteful's trial is locked behind Idenfy KYC. You need a device with a camera: a Mac mini or a PC without one cannot complete it. I tried, hit the wall, switched to a MacBook with a camera, and Idenfy responded with "Token does not exist" / "REPO_MISSING_TOKEN" and a link to Idenfy's documentation. No retry explanation, no path forward. I've done enough KYC flows to know: this one blocked midway for no reason I could see.

The same KYC reappears if you want to pay in crypto: "need more KYC details to do the payments." So the free trial and the crypto option are both gated behind the same identity wall that failed me.

My soft-observations verdict: everything with Byteful is deceiving: pretty on the surface, a locking mechanism to move forward. Yay, a 1GB free trial: locked with KYC. Yay, crypto payments: locked with more KYC. Okay, let me verify: it blocks midway. Every step, a lock.

The dashboard that hides its prices

Inside, the dashboard is clean and organized: it looks like webshare at first glance. Active services, usage stats, online IP counts, a live activity page for debugging. That part is genuinely fine.

The problem is buying. Every package's price, static, residential, mobile, datacenter, is hidden behind a button. In the dashboard, you cannot see what you're purchasing by clicking "Purchase now". The only way to see prices is to leave the dashboard and go back to the public pricing page. That's not a feature; it's a UX decision that makes you spend money blind.

The entry price is a fiction in practice. The residential package starts at $3.25/GB, but you can only buy a package with account balance, and the minimum balance load is $5. Trying to add $3.50 or $4.00 throws an error. So the "cheapest test" is actually $5 loaded, and the smallest real purchase is the 2GB/$6.50 package I bought. The marketing says "no monthly minimum purchase requirements": my experience says there's a $5 floor on every purchase.

Scaling is linear and pricey: 10GB is $32.50, the $3.25/GB rate holds with zero volume discount at the entry tiers. 25GB drops to $3/GB (8% off), 100GB to $2.75/GB (15% off). Compare webshare, where the entry point is $3.50 with no balance gimmick. Byteful's one genuine positive here is crypto acceptance (Bitcoin, Ethereum, Solana, Monero, Binance Coin, Tether), until the KYC clause appears.

The OS filter nobody else has

One thing is genuinely unique: OS selection on residential proxies. You can combine rotating country/state/city filters with an OS filter (Windows, macOS, mobile), or go country-city, country-asn, country-zip for more targeted IPs. That combination, OS plus geo, is only available on Byteful among the six I tested. Dataimpulse has deeper filters (country-city-zip-asn) but no OS dimension.

It's a real control point for anyone mimicking device behavior. The trade-off is the usual one: more filters, fewer IPs. My pool drain came back 148/149 unique IPs (99.3% rotation, first repeat at request 73), and a note from the OS+geo combination is that the pool shrinks as you narrow it: normal industry behavior, worth planning around.

Support that asks for your name

Support is email-only. No chat bot: in 2026, a trained chatbot is the industry standard. Webshare's Webby answered everything in real time (it even admitted their landing page was misleading for my use case); decodo has Codette. Byteful emails you back.

My direct experience: I signed in, went to support, and the flow asked me "what's your name?" and for my email, from a dashboard where I was already signed in. It reads like a tech glitch, but it's an organizational issue: disconnected components asking for data they already have. The company markets "interact with support" from the dashboard and lists live chat, email, tickets, Discord, Telegram: on my tested experience, it was email, and the conversation restarted its context every time.

The API that manages, and the Smartpath I didn't test

Byteful doesn't sell a web scraping API (send a URL, get data). What it has is a REST management API: 40+ operational endpoints covering nearly everything the dashboard does (buying plans, creating and managing users, retrieving pool and usage information) with a 10 requests/second rate limit for all users. Marketing says the docs are "the most detailed in the industry." I did not benchmark the API; treat that as on-paper.

Smartpath is the feature their marketing leans on hardest: an AI router that shifts non-essential requests to cheaper datacenter nodes to save bandwidth. Marketing claims sub-0.5-second connection times and 30-60% bandwidth savings, and says "requests routed through Smartpath cost nothing." A third-party review found it saved 30MB over 500 requests in a small test, but also that it's not ideal for ad verification, multi-accounting, or sensitive fingerprinting, because it shifts IP classifications mid-session. I did not run Smartpath. No numbers from me: but if you use it, be aware it is designed to change your exit-node profile behind your back.

SERP tracking: not tested yet

No dedicated SERP benchmark on Byteful yet: I'll update this section when it lands. Product signals only: Byteful markets SERP proxies for "collecting accurate, localized search results," which is a proxy-for-SERP promise rather than a parsed SERP API. A third-party review says Byteful blocks SERP scraping on Google when using its static proxy networks, which conflicts with the SERP-proxy marketing, and I can't resolve it from my test plan. Treat SERP as untested until I run it.

The limits: fast, cheap entry, then a wall

Cost per 1k requests: $0.46, mid-tier of the six (dataimpulse $0.15, shifter $0.33, webshare $0.51, decodo $0.63, oxylabs $0.86). Never-expiring traffic on residential bandwidth is real, and the 1GB free trial is the most generous cap of the set: both are honest wins.

The walls on entry: $5 minimum balance to buy anything, Idenfy KYC for the trial and for crypto, pricing hidden behind buttons. The refund policy is thin but defined: full refund within 24 hours if under 2GB or 25% of allocation, once per lifetime, excluding free trials and custom contracts.

Concurrency ceiling: c8, same as everyone except dataimpulse's c16. Marketing says "unlimited concurrent sessions without performance degradation"; I measured the ceiling, not the promise. Usage policy blocks government sites, banking/financial services, payment processors, mailing, NSFW, and some e-commerce stores (per third-party review), and you can't delete your account yourself.

Untested gaps: SERP tools, Smartpath, sustained 5-minute throughput runs, home-IP baseline lift, mobile/ISP/dedicated-datacenter tiers, and ASN/country pool diversity (ipinfo backfill pending).

The verdict: fast nodes, hard walls, hidden onboarding

Byteful is the fastest option I measured for high-volume scraping of easy, US-targeted sites: 961ms median, 1.61 RPS, a US gateway at 750ms (the fastest of all six), the only US-heavy default pool (34% US egress, next highest 19%), and the only pool that gets faster under load. If your targets are agoda/amazon/imdb/ozbargain-class and you need throughput, the numbers favor Byteful. Same for slickdeals, 26%, best of the set. And if you need OS-level proxy targeting, no one else offers it.

Not for: zillow, ebay, or any PerimeterX-adjacent target (0% and 5% are the worst results of all six). Not for anyone who values onboarding or transparency: hidden pricing, $5 minimum, camera-locked KYC. Not for privacy-sensitive users: Idenfy KYC is mandatory, not optional. Not for scaling budgets: $3.25/GB holds to 10GB with no discount, and every purchase needs a $5 balance floor.

Alternatives

If Byteful's walls or onboarding don't fit, the honest alternatives from the same test set:

• Decodo: $0.63/1k, 76.9% success, 1,191ms, 0.1% timeouts. The reliable-transport pick.
• DataImpulse: $0.15/1k, 76.8%, the only c16 ceiling, zillow 95%. The budget-and-concurrency pick.
• Webshare: $0.51/1k, 1,242ms, plan customization, the best chatbot I tested. The friendlier mid-tier pick.
• Shifter: $0.33/1k, most diverse pool (117 ASNs), zillow 96%, but 2,402ms and 4.5% timeouts.

Buy Byteful for speed on easy targets, never for tough ones, and budget the onboarding time before you buy. The speed is real and it scales. The walls are real, and fast doesn't mean successful. If the exit-node reputation and the onboarding get fixed, this becomes a different review: I'll update it when I retest.