Roundup
The best mobile proxies, and when you don't need one
I compared 9 mobile proxy providers on what you pay at 10 and 100 GB, Proxyway's genuine-mobile-IP data and refund terms. Plus when mobile is overkill.
For a mobile pool that is really mobile at scale, I’d pick Oxylabs. For the best value with a way out, Decodo. If genuine carrier IPs and a long refund window matter most, Evomi. I compared nine providers’ pricing pages and terms on November 12, 2026, and set them against Proxyway’s published 2026 mobile test.
Two numbers decide most mobile purchases: how much of the pool really is carrier IPs, and how much cash you commit before the first request. The pages ranking for this query mostly skip both. If you’re after Google results, jump to the Google section first.
Short on time?
Here's who I'd pick
- Really-mobile pool at scale
- Oxylabs
- Top success rate (99.84%) and the most unique IPs in Proxyway's mobile test. $500 buys 100 GB; self-service purchases aren't refundable.
- Best value, with a refund window
- Decodo
- $275 for 100 GB on its current 50%-off sale, 99.00% genuine mobile IPs on average in Proxyway's test, and 14 days to get your money back on a first purchase.
- Most genuinely mobile, longest refund
- Evomi
- Highest average genuine-mobile share (99.29%) in Proxyway's test and a 30-day refund on monthly plans. 10 GB is $39.80 pay-as-you-go.
- Traffic that never expires
- Byteful
- Proxyway called it the fastest mobile network. $42.50 for 10 GB or $345 for 100 GB, and unused GB don't lapse.
- Traffic in cheaper countries
- SOAX
- Priced by exit country: $5/GB for the US on pay-as-you-go, down to $0.35/GB for Tier 3 countries on the largest plan.
- Smallest first purchase
- DataImpulse
- $5 for 2.5 GB if you're new, then $2/GB; only NodeMaven's $3.50 trial costs less up front. Proxyway found only 78.36% genuine mobile IPs on average, so check your country.
- Named carriers or a port by the day
- IPRoyal
- 13 carriers shown by name and dedicated ports from $10.11 a day. No independent test data on its pool yet.
- Google search results
- A SERP API, not a mobile proxy
- Every raw mobile pool in ScrapingBee's public test scored 0% on usable Google results.
Prices verified . Sponsors cannot buy a verdict or a position.
Do you actually need a mobile proxy? #
A mobile proxy sends your requests out through a phone-network connection, so the site sees a carrier IP. Carriers share a small set of public addresses across many subscribers, so blocking one of those IPs can block real customers too. Vendor explainers such as Byteful’s FAQ make the same argument.
You need mobile when the job is about the mobile network itself: ad or app checks on a specific carrier, content shown only to mobile visitors, or carrier-level geography. For most scraping, residential or ISP IPs do the work for less, and datacenter IPs cost less again. My guides to residential vs datacenter and ISP vs residential cover those choices.
The usual advice says mobile costs two to three times residential. That only holds for some vendors, so I put each one’s 100 GB rates side by side (verified 2026-11-12).
| Provider | Mobile, $/GB at 100 GB | Residential, $/GB at 100 GB | Gap |
|---|---|---|---|
| Decodo | $2.75 | $2.75 | None |
| SOAX | Same rate | Same rate | “One rate across residential and mobile” |
| NodeMaven | One plan | One plan | Shared traffic |
| Oxylabs | $5 | $4 | 1.25× |
| Byteful | $3.45 | $2.75 | 1.25× |
| DataImpulse | $2 | $1 | 2× |
| Evomi | $3.20 (monthly) | $0.49 Core, $3.98 Premium PAYG | Depends on pool |
My rule: if your vendor charges the same for both, price is no longer the argument against mobile, and the only question is whether you need carrier IPs. Where mobile costs more, buy residential first and move up only if a carrier IP is the point of the job.
How do the nine providers compare? #
Cash means what you pay at checkout to get at least that many GB, in USD before tax. Genuine-mobile share is Proxyway’s figure averaged across seven test locations.
| Provider | Best for | Cash for 10 GB | Cash for 100 GB | Genuine mobile (Proxyway, 7-location avg) | Sticky cap | Carrier targeting | Refund |
|---|---|---|---|---|---|---|---|
| Oxylabs Top pick | Really-mobile pool at scale | $100 (Basic plan) | $500 | 99.04% | 24 h | Claimed, carriers not named | None on self-service |
| Decodo Best value | Best value with a refund window | $40 (PAYG) | $275 (50%-off sale; $550 list) | 99.00% | Not stated for mobile | Not stated; 700+ carriers in pool | 14 days, first purchase, 20% use cap |
| Evomi | Most genuinely mobile | $39.80 (PAYG) | $320 (monthly) | 99.29% | 24 h | Yes | 30 days on plans, not PAYG |
| SOAX | Cheaper exit countries | $50 (US/DE) | $500 (US/DE) | 99.28% | Not stated | Not stated | Generally non-refundable |
| Byteful | Fast, never expires | $42.50 | $345 | 98.39% | 24 h | Not named | Not stated |
| DataImpulse | Smallest first purchase | $50 (25 GB pack) | Check at checkout ($2/GB rate) | 78.36% | 120 min | ASN at extra cost | 7 days, first Intro only |
| IPRoyal | Named carriers, day-rate ports | $60 | $520 | Not tested | Not stated | 13 named carriers | 24 h, under 100 MB used |
| NodeMaven | One plan for residential and mobile | No 10 GB stop (8 GB $32, 20 GB $75) | $300 (monthly) | Not tested | 24 h | Not stated | Not on pricing page |
| Proxidize | Flat $2/GB after KYC | $50 (25 GB minimum) | $200 | Not tested | 3 days | Not stated | 30 days, under 10 GB used |
Source: Each provider's mobile pricing and terms pages, read 12 Nov 2026; genuine-mobile share from Proxyway Proxy Market Research 2026. Scroll sideways for more columns.
Which pools are really mobile? #
A “mobile” pool can include IPs that aren’t on a carrier network at all. Proxyway’s 2026 market research counted an IP as genuinely mobile when IP2Location gave it a mobile network code. It sent about 1.2 million requests per provider to each global pool, mostly in March and April 2026.
Three caveats before the numbers. Providers saw the method in advance, and Proxyway says the results reflect the “best rather than default configuration”. The page also gives the test length as 21 days in the text but 14 days in a table.
Proxyway earns affiliate fees from providers, and its page says so: “We use affiliate links.” I quote its figures as its own and name the column each comes from.
| Provider | Genuine mobile, avg of 7 locations | Genuine mobile, global pool | Success rate | Response time |
|---|---|---|---|---|
| Evomi | 99.29% | 98.63% | 99.66% | 0.93 s |
| SOAX | 99.28% | 97.64% | 99.74% | 0.92 s |
| Oxylabs | 99.04% | 99.55% | 99.84% | 0.76 s |
| Decodo | 99.00% | 99.50% | 99.75% | 0.74 s |
| Byteful | 98.39% | 97.18% | Not listed | 0.48 s (fastest) |
| DataImpulse | 78.36% | 86.97% | 99.69% | 1.00 s |
| NetNut (seized, reference only) | 75.14% | 79.07% | 95.66% | 1.76 s |
Source: Proxyway Proxy Market Research 2026, read 2026-11-12. Proxyway’s mobile success table doesn’t list Byteful. IPRoyal, NodeMaven and Proxidize weren’t in its mobile test.
The column you pick changes the story. On the seven-location average, Evomi and SOAX lead; on the global pool, Oxylabs and Decodo do. DataImpulse is the one big outlier either way, and its share swings by country: 95.78% in the US, 80.86% in the UK, 71.58% in Australia and 29.69% in Brazil.
Genuine-mobile share isn’t the same as getting through, though. On Proxyway’s real-target run against Amazon, Google and Instagram, the mobile median was 75.26%, and DataImpulse posted the best result at 85.31%. Decodo was fastest on those targets at 2.71 s.
Unique IPs tell you how much of the pool you’ll actually see. Oxylabs returned 1,192,449 unique IPs from 1.2 million requests, the most of any provider, and Decodo returned the most unique US IPs (152,424 of 280,000 requests). SOAX “returned the fewest”, and Proxyway notes it assigns IPs close to the requester.
Oxylabs: the safest bet at scale #
Best for teams committing $500 a month or more that want a pool independent data says is really mobile. Proxyway measured the top success rate (99.84%), the most unique IPs and 99.55% genuine mobile on the global pool. In ScrapingBee’s test, its raw pool hit 100% on Amazon and Instagram.
Prices on its mobile page (verified 2026-11-12): $7.5/GB up to 2 GB on the $15 Starter plan, $6.67/GB on the $100 Basic plan (listed for up to 14 GB), $5/GB on the $500 Flexible plan (up to 100 GB) and $3.5/GB from 714 GB on Enterprise, from $2,500. The plan is one credit wallet shared across Oxylabs products, and the page advertises a $5 welcome credit. How that wallet works is in my Oxylabs pricing breakdown.
The pool claim is 20M+ IPs in 140+ countries, with 4,429,824 in the US, 1,449,139 in the UK and 1,431,960 in Germany. Targeting runs down to city, coordinates and ASN, and carrier targeting is listed at no extra charge, though no carriers are named. Sessions stick for up to 24 hours over HTTP, HTTPS, HTTP3 or SOCKS5.
The drawback is the terms. The page says self-service purchases are non-refundable, unused credits expire at the end of each billing cycle, and traffic is billed “regardless of the HTTP status”, so failed requests cost you. Its general conditions (last updated September 25, 2024) allow a pay-as-you-go refund under six conditions, which conflicts with the lander, so ask before you pay.
Decodo: the best value, with a refund window #
Best for most buyers who want a really-mobile pool and a way out if it doesn’t fit. Proxyway measured 99.75% success, a 0.74 s response (the fastest in its success table) and 99.00% genuine mobile on average. It was also fastest on real targets.
The mobile pricing page lists monthly plans of 2 GB for $7.5, 8 GB for $28, 25 GB for $81.3, 50 GB for $150, 100 GB for $275, 250 GB for $625 and 500 GB for $1,125 ($2.25/GB), all marked “SAVE 50%” (verified 2026-11-12).
Pay-as-you-go shows $4/GB with $8 struck through, yet the FAQ on the same page says 1 GB pay-as-you-go is “priced at $8”. The page also says a subscription and pay-as-you-go “cannot be used at the same time”, so you can’t top up a monthly plan with PAYG GB.
The pool claim is 10M+ IPs, 700+ carriers and 160+ locations, with country and city targeting included. It runs over HTTP(S) and SOCKS5. The full price ladder sits in my Decodo pricing guide.
Terms: the residential pricing page describes a 3-day, 100 MB trial that turns into the paid plan unless you cancel. It also offers a 14-day money-back window on a first purchase if you didn’t use the trial and used no more than 20% of the traffic.
The mobile page mentions a free trial and a “14-day money-back option” but doesn’t restate the conditions, and the licence says credits expire after 12 unused months.
The drawback is the $4 versus $8 conflict on its own page; budget for $8 until checkout shows otherwise. Decodo and Oxylabs are sister companies in the Source group, announced on October 13, 2026, which says each brand runs independently.
Evomi: the most genuinely mobile pool Proxyway measured #
Best for buyers who want the highest genuine-mobile share and the most forgiving refund terms here. Proxyway measured 99.29% genuine mobile on average (the highest), 99.66% success and 0.93 s. It also recorded Evomi’s US results as the best in the mobile test: 99.95% success and a 0.41 s response.
On its mobile page (verified 2026-11-12), pay-as-you-go is $3.98/GB and doesn’t expire, per its pricing page. Monthly plans run from 8 GB for $30 and 25 GB for $90 through 50 GB for $170, 100 GB for $320 and 250 GB for $750, up to 5,000 GB for $11,000 ($2.20/GB).
The pool claim is 3M+ IPs, the smallest headline here. You target a carrier with an isp- parameter, sessions stick for up to 1,440 minutes (30 by default), and it supports HTTP, HTTPS and SOCKS5.
The refund policy gives 30 days on your first purchase in a product category, while you’ve used under 1 GB or 10%, whichever comes first. Pay-as-you-go and crypto payments are excluded, so the refund covers monthly plans, not the $3.98 route. That’s the drawback: the cheapest way to 10 GB is the one you can’t get back.
SOAX: cheapest when your traffic exits in cheaper countries #
Best for traffic that exits in lower-cost countries, or for buyers mixing residential and mobile. Proxyway measured 99.74% success, 0.92 s and 99.28% genuine mobile on average (97.64% on the global pool).
SOAX pricing uses credits priced by geo tier and “one rate across residential and mobile” (verified 2026-11-12). Sandbox has no monthly fee and charges $5, $4 or $2 per GB for Tier 1, 2 or 3 countries. Builder is $200 a month ($3/$2.25/$1.20), Team $500 ($2.20/$1.65/$0.90), Scale $1,500 and Enterprise $3,000 (from $0.85/$0.55/$0.35); the FAQ places the US and Germany in Tier 1 but doesn’t say where the UK sits.
Rotation can run per request, every N requests, by time or on error; the sticky cap isn’t stated. Pool claims conflict: the mobile page says “over 191 million residential and mobile IPs” while the residential page says “over 12 million”.
The drawback is the terms. There’s no free trial, credits expire after 60 days (12 months on annual plans) and payments are “generally non-refundable”. Proxyway also found SOAX returned the fewest unique IPs, so it may not suit jobs that need many distinct addresses.
Byteful: fast, and the traffic never expires #
Best for mid-volume buyers who want speed and traffic that doesn’t lapse. Proxyway wrote that Byteful “took the crown as the fastest mobile proxy network”, at 0.48 s on the global pool, with 98.39% genuine mobile on average.
Its mobile page lists 10 GB at $4.25/GB ($42.50), 25 GB at $3.85 ($96.25) and 100 GB at $3.45 ($345), plus custom deals (verified 2026-11-12). The title and schema advertise “$2.25/GB”, but no visible tier is that low. Traffic never expires.
The pool claim is 6M+ IPs in 195+ countries and “1,000+” carriers, none named. Its monthly active counts (586,161 US, 326,045 UK, 296,049 Germany) are the same figures its residential page shows, which makes me read them as network-wide rather than mobile-only. Sessions stick for 24 hours over HTTP, HTTPS or SOCKS5, and IPs are “sourced through verified SDK partners”.
The drawback: there’s no mobile trial, and I couldn’t find a refund policy. If you can’t get money back, start with the 10 GB tier.
DataImpulse: the lowest entry price, with a pool caveat #
Best for the smallest first purchase, mainly for US traffic. Only NodeMaven’s $3.50, 750 MB trial costs less up front. The mobile page offers an Intro pack of 2.5 GB for $5 to new users, then Basic at $50 for 25 GB ($2/GB), Advanced at $1,600 for 1 TB ($1.60/GB) and Custom from $8,000 for 5 TB+ (verified 2026-11-12). Traffic never expires.
The pool claim is 16M+ IPs, with 111,231 in the US. Country targeting is included, while city, ZIP and ASN carry “extra costs”. Sticky sessions last up to 120 minutes, and the pool lists 3G, 4G, 5G and LTE.
Proxyway’s results cut both ways. It measured 99.69% success but only 78.36% genuine mobile on average, with 29.69% in Brazil and 95.78% in the US. It also gave DataImpulse the best real-target score (85.31%), so the non-mobile IPs don’t seem to stop it getting through.
The drawback is the refund. Section 10.1 of the terms limits the 7-day refund to a first-time purchase of the Intro plan, and crypto payments are excluded. Trustpilot has also withheld DataImpulse’s TrustScore over a guideline breach.
IPRoyal: named carriers and day-rate ports #
Best for jobs that need a named carrier, or a dedicated mobile port you rent by the day. IPRoyal shows 13 carrier logos on its mobile page: T-Mobile, Verizon, Orange, Free, Vodafone, Three, W3, A1, Maxis, Yes, Telia, Bite and Optus. No other provider here names that many.
Rotating traffic costs $6.80/GB for 2 GB ($13.60), $6 for 10 GB ($60), $5.60 for 50 GB ($280) and $5.20 for 100 GB ($520) (verified 2026-11-12). A dedicated port is $10.11 a day or $130 for 30 days, dropping to $117 a month on a 90-day term, with a 30 GB daily cap per port. It claims 4.5M+ IPs, 15 countries with roughly 470K to 521K IPs each, unlimited concurrent sessions, and HTTP, HTTPS and SOCKS5.
There’s no free trial, only a paid 24-hour dedicated test plan. Refunds or replacements come within 24 hours if you’ve used under 100 MB (rotating) or 500 MB (dedicated), and the terms allow a service fee at payment (§6.17) and exclude crypto refunds (§6.9.4).
The drawback: IPRoyal wasn’t in Proxyway’s mobile test, so there’s no independent read on how mobile its pool is. Trustpilot has also withheld its TrustScore over a guideline breach.
NodeMaven: one plan for residential and mobile #
Best for buyers who want residential and mobile from one balance, with unused traffic rolling over. Its pricing page sells one rotating plan covering “30M+ Residential + 250K Mobile IPs” from shared traffic (verified 2026-11-12).
Monthly billing is 8 GB for $32 ($4/GB), 20 GB for $75 ($3.75) and 100 GB for $300 ($3). Pay-as-you-go is 8 GB for $39, 20 GB for $90 and 100 GB for $345; the slider has no 10 GB stop. A trial costs $3.50 for 750 MB, sessions stick for up to 24 hours, and the card lists traffic rollover.
The card also claims 99.54% average success, under 0.3 s and a 95% clean-IP rate; those are NodeMaven’s own numbers. Independent data is limited to ScrapingBee’s test, where its raw pool scored 0% on Google, 100% on Amazon and 99% on Instagram.
The drawback: mobile is 250K IPs inside a much bigger residential plan, and the pricing page doesn’t state refund terms.
Proxidize: flat per-GB or per-proxy, after KYC #
Best for verified buyers who want one flat rate at volume, or a fixed mobile proxy billed monthly. Its pricing page describes a hosted service with “No hardware or DIY setup”, so you aren’t buying devices.
Mobile traffic is $2/GB flat, from $50 for 25 GB up to 10 TB, or $59 per proxy per month (verified 2026-11-12). Bandwidth expires after a month, and every buyer goes through KYC. Its mobile page lists sticky sessions of up to 3 days and UDP over SOCKS5.
You get a 30-day money-back guarantee if you’ve used under 10 GB. The drawback is that nobody has published independent test data on its pool, and its “ethically sourced” claim doesn’t explain how device owners opt in.
One footnote: Rayobyte also sells mobile, and Proxyway measured 98.22% genuine mobile on average. Its smallest mobile plan was $250 for 83 GB, expiring after 30 days, per its pricing page, which is too much commitment for a first purchase.
So what do mobile proxies cost? #
At 10 GB you’ll pay between about $40 and $100 in cash; at 100 GB, between $200 and $520. The spread comes from minimum plans more than from the per-GB rate. All figures are USD before tax, verified 2026-11-12.
| Provider | What you buy for 10 GB | Cash | What you buy for 100 GB | Cash |
|---|---|---|---|---|
| Evomi | PAYG, 10 × $3.98 | $39.80 | Performance plan, 100 GB a month | $320 |
| Decodo | PAYG, 10 × $4 ($80 if the FAQ’s $8 applies) | $40 | 100 GB monthly plan | $275 on the 50%-off sale ($550 list) |
| Byteful | 10 GB tier | $42.50 | 100 GB tier | $345 |
| DataImpulse | Basic, 25 GB (15 GB left over, never expires) | $50 | $2/GB rate | Check at checkout |
| SOAX, US or Germany | Sandbox top-up, 10 × $5 | $50 | Sandbox top-up, or the Team plan (about 227 GB at $2.20) | $500 |
| Proxidize | Minimum 25 GB, expires in a month | $50 | 100 GB at $2 | $200 |
| IPRoyal | 10 GB tier | $60 | 100 GB tier | $520 |
| Oxylabs | Basic plan at $6.67/GB | $100 | Flexible plan at $5/GB | $500 |
| NodeMaven | No 10 GB stop: 8 GB $32 or 20 GB $75 a month | n/a | 100 GB monthly ($345 PAYG) | $300 |
For context, Proxyway’s 2026 report puts the median mobile entry price at $30, with medians of $4.38/GB at 5 GB, $3.50 at 50 GB and $2.48 at 500 GB.
Per-port pricing is a different product, so I’ve kept it out of the table. IPRoyal rents a dedicated mobile port for $10.11 a day or $130 for 30 days, and Proxidize charges $59 per proxy per month.
Which billing rules catch people out? #
- Oxylabs bills failed requests. Traffic counts “regardless of the HTTP status”, and unused credits lapse at the end of each cycle. Oxylabs
- Decodo’s PAYG price conflicts on its own page, $4 in the table and $8 in the FAQ, and you can’t mix PAYG with a monthly plan. Decodo
- SOAX’s price depends on the exit country, and credits expire after 60 days. SOAX
- Short expiry windows: Proxidize bandwidth expires monthly and Rayobyte’s after 30 days, while Byteful, DataImpulse and Evomi PAYG traffic doesn’t expire.
Rotating, sticky or dedicated? #
Rotating gives you a new IP per request or on a timer, which suits broad collection. Sticky keeps one IP for a session, which you need when a task spans several steps. Dedicated means a port that’s yours alone, billed by time rather than GB.
Most providers here cap sticky sessions at 24 hours: Oxylabs, Byteful, NodeMaven and Evomi (1,440 minutes). DataImpulse stops at 120 minutes and Proxidize goes to 3 days. Decodo, SOAX and IPRoyal don’t state a mobile cap, so ask before you rely on a long session.
For a dedicated port, IPRoyal and Proxidize are the two that sell one openly. Pick IPRoyal for short jobs billed by the day, and Proxidize for a month or more.
Does 3G, 4G or 5G matter? #
I couldn’t find independent data showing that a newer radio generation gets blocked less. Oxylabs and DataImpulse list 3G, 4G, 5G and LTE in one pool rather than selling them separately.
What changes outcomes in the data I read is the carrier, the country and how much of the pool is really mobile. I’d choose on those and ignore “5G” in a product name.
Can you target a city or a carrier? #
Country targeting is standard. City, ASN and carrier targeting vary, and sometimes cost extra: DataImpulse charges for city, ZIP and ASN, while Oxylabs lists carrier targeting at no extra charge. Headline pools also say little about the country you need, so here are the vendor-stated counts for three common markets.
| Provider | Headline pool | US | UK | Germany | Named carriers |
|---|---|---|---|---|---|
| Oxylabs | 20M+ | 4,429,824 | 1,449,139 | 1,431,960 | None named |
| Byteful | 6M+ | 586,161 | 326,045 | 296,049 | None named |
| IPRoyal | 4.5M+ | About 470K to 521K per listed country | 13 | ||
| DataImpulse | 16M+ | 111,231 | Not stated | Not stated | Not stated |
| Decodo | 10M+ | Most unique US IPs in Proxyway’s test (152,424) | Not stated | Not stated | 700+ claimed, not listed |
Byteful’s counts are monthly active IPs and match its residential page; Oxylabs doesn’t say how it counts. Treat these as vendor claims, not measurements, and compare them only within one vendor.
Trials, refunds and expiry, side by side #
This is the table I’d read before paying, because a few of these vendors give you no way back. Terms verified 2026-11-12.
| Provider | Trial | Refund | Traffic expiry | Billing note |
|---|---|---|---|---|
| Oxylabs | $5 welcome credit | Self-service “non-refundable” (general conditions differ) | End of each billing cycle | Billed regardless of HTTP status |
| Decodo | 3 days, 100 MB; becomes the paid plan | 14 days, first purchase, 20% use cap, no trial used | Credits after 12 unused months | PAYG and monthly can’t run together |
| Evomi | Advertised, terms not stated | 30 days, first purchase, under 1 GB or 10%; not PAYG or crypto | PAYG never expires | Monthly plans |
| SOAX | None | “Generally non-refundable” | 60 days (12 months annual) | Price set by exit country |
| Byteful | None for mobile | Not stated | Never | |
| DataImpulse | Intro pack ($5) | 7 days, first Intro purchase only; not crypto | Never | City, ZIP, ASN extra |
| IPRoyal | Paid 24 h dedicated test | 24 h, under 100 MB rotating or 500 MB dedicated | Not stated | Possible service fee (§6.17) |
| NodeMaven | $3.50 for 750 MB | Not on pricing page | Unused traffic rolls over | |
| Proxidize | None listed | 30 days, under 10 GB used | 1 month | KYC for every buyer |
If you want to try before committing, Evomi’s monthly plans and Decodo’s first purchase give you the most room. If refunds don’t matter but expiry does, Byteful and DataImpulse are the two whose traffic never runs out.
Can websites tell it’s a mobile proxy? #
Yes, they can usually tell the IP belongs to a mobile carrier, because IP databases label it. That’s the same label Proxyway used to measure genuine-mobile share, via IP2Location’s mobile network code. What a carrier IP buys you is reluctance to block, not invisibility.
Sites still block on behaviour. On Proxyway’s real-target run, mobile pools got through on Amazon, Google and Instagram a median 75.26% of the time, against near-99% success on its plain test endpoint. Request rate, headers and browser fingerprint matter as much as the IP.
Setting one up takes two minutes #
A mobile proxy works like any other proxy: you get a gateway host, a port and credentials, and you paste them into your HTTP client, browser or scraping tool. Oxylabs, Decodo, SOAX, Evomi, Byteful and IPRoyal list both HTTP(S) and SOCKS5; Oxylabs adds HTTP3, and Proxidize lists UDP over SOCKS5.
Rotation and targeting usually ride in the credentials. Evomi, for example, uses an isp- key for carrier and a session setting with a 30-minute default. Check each provider’s docs for the exact syntax rather than copying another vendor’s format.
What happened to Bright Data’s and NetNut’s mobile pools? #
Two names that older lists still rank shouldn’t be on your shortlist today.
Bright Data no longer lists or prices a mobile product. On November 12, 2026, its mobile proxies URL landed on a general proxy page listing residential, datacenter and ISP, and its mobile pricing URL returned a 404. Its own release notes from October 24, 2026 retired mobile proxies for “immediate access” accounts, and said retired proxies would be deleted on November 9, 2026.
NetNut’s public domain, netnut.io, was still headed “Seized by the Federal Bureau of Investigation” when I checked on November 12, 2026. That is an observed page state; authenticity, the underlying court order and legal outcome were not independently established in this research. If you were a NetNut customer, NetNut alternatives covers where to move.
KrebsOnSecurity reported on July 2, 2026 that the FBI and IRS-CI had seized NetNut domains tied to the Popa botnet of “at least two million devices”, and that Alarum said it would “fully cooperate”. Google’s threat intelligence team wrote a day later that Play Protect had disabled apps carrying NetNut SDKs.
Where do mobile IPs come from? #
Mostly from apps on real phones. A vendor’s SDK runs inside an app, and the phone owner agrees, or should agree, to share bandwidth. The FBI’s March 2026 notice warns that devices can join these networks with or without the owner’s consent, and that SDK makers pay app developers to carry them.
Here is what each vendor says, quoted as their claim:
- Oxylabs: “ethically procured” (mobile page)
- Byteful: “sourced through verified SDK partners” (mobile page)
- Evomi: “paid, opt-in volunteer nodes” (mobile page)
- Proxidize: “ethically sourced” and opt-in, with no mechanism described (mobile page)
None of them publishes an independent consent audit. Before you buy, I’d ask four things: which apps carry the SDK, what the user sees when they opt in, what they’re paid, and how they opt out. My ethical residential proxies guide has the full checklist, and it applies to mobile pools too.
Are mobile proxies legal, and what about Instagram or TikTok? #
Using a proxy isn’t usually the legal question. What you collect, from which site and under whose terms is. A proxy changes your IP; it reduces blocks, but it doesn’t grant permission or override a platform’s rules.
That matters most on social platforms. Running accounts or automation on Instagram or TikTok through mobile IPs is a question for those platforms’ terms and your own agreements with them, so I’d clear the use there before buying traffic. For the general picture, see my guide to whether web scraping is legal.
Scraping Google? Don’t use mobile proxies #
If your target is Google search results, mobile proxies are the wrong tool. ScrapingBee’s public test sent 400 requests per provider through Italian mobile IPs and counted a success only when real content came back. In its words, “Every conventional mobile proxy provider scored 0% on usable Google SERPs”, while Amazon and Instagram came in near 100%.
Weigh the source. The author co-founded ScrapingBee, which sells a scraping API and calls its own service “the strongest candidate from this benchmark” for heavily protected targets, and ScrapingBee belongs to the same Source group as Oxylabs and Decodo, two of the pools tested. The test dates aren’t published, but the raw data is on GitHub, and it’s the only reproducible mobile test I found among the top results.
For Google results, a SERP API returns parsed results for a per-search price, and you don’t manage IPs at all. My SERP API comparison covers the options.
Which one should you pick? #
- Really-mobile pool at scale: Oxylabs, if $500 a month and no self-service refunds are acceptable. Runner-up: Decodo.
- Best value with a way out: Decodo, at $275 for 100 GB with a 14-day money-back window, while its 50%-off sale runs. At the $550 list rate it is the dearest here, so check the price at checkout; if the sale has ended, Evomi ($320, 30-day refund) takes this slot. Budget for its $8 FAQ price on PAYG until checkout says otherwise.
- Genuine carrier IPs and the longest refund: Evomi, on a monthly plan so the 30-day refund applies.
- Traffic that never expires: Byteful, starting with the 10 GB tier since its refund terms aren’t published.
- Cheaper exit countries: SOAX, if you can use the credits within 60 days.
- Smallest first purchase: DataImpulse’s $5 Intro, best for US traffic given its genuine-mobile numbers elsewhere.
- A named carrier or a port by the day: IPRoyal.
- Residential and mobile from one balance: NodeMaven.
- Google results: a SERP API, not a proxy.
Still deciding between proxy types? My roundup of proxy providers for web scraping picks one per job, my residential proxy picks cover the cheaper tier, and the proxies hub maps the rest.
FAQ
Are mobile proxies worth it over residential?
Only when the job needs a carrier IP, such as checking ads or content as a mobile user on a given network. For most scraping, residential or ISP IPs do the same job. Where a vendor charges the same per GB for both (Decodo at 100 GB, SOAX, NodeMaven), price stops being the reason to avoid mobile.
How much do mobile proxies cost?
On November 12, 2026, 10 GB cost between $39.80 (Evomi pay-as-you-go) and $100 (Oxylabs, whose smallest plan covering 10 GB is $100). 100 GB ran from $200 (Proxidize) to $520 (IPRoyal), with Decodo at $275 on its current 50%-off sale ($550 at the struck-through list rate). Proxyway's 2026 report puts the median at $4.38/GB for 5 GB and $3.50/GB for 50 GB.
Can websites detect mobile proxies?
Sites can see that an IP belongs to a mobile carrier, because IP databases label it. That is how Proxyway measured genuine-mobile share. They can still block you on behaviour: Proxyway's real-target run on Amazon, Google and Instagram had a 75.26% median success rate for mobile pools.
Do I need 5G mobile proxies?
I found no independent data showing that 5G beats 4G for unblocking. Vendors such as Oxylabs and DataImpulse list 3G, 4G, 5G and LTE in one pool. Carrier and country matter more for most jobs than the radio generation.
Rotating or sticky mobile proxies?
Rotate for broad data collection, where each request can come from a new IP. Use a sticky session when a task needs the same IP across several steps; most providers here cap it at 24 hours, DataImpulse at 120 minutes and Proxidize at 3 days. For a fixed IP you control, IPRoyal and Proxidize rent dedicated ports.
Do mobile proxies work for scraping Google?
Not on their own. In ScrapingBee's public test, every raw mobile proxy pool scored 0% on usable Google results, while Amazon and Instagram were near 100%. For Google results, a SERP API is the better tool.
Is Bright Data still selling mobile proxies?
Bright Data no longer lists or prices a mobile product: its mobile URL lands on a general proxy page and its mobile pricing page returns a 404. Its own release notes from October 24, 2026 retired mobile proxies for immediate-access accounts, with deletion on November 9, 2026.
Are mobile proxies legal?
Buying and using a proxy is not the legal question. What you collect, from which site, under whose terms, is. A proxy changes your IP; it does not give you permission to access a site or override a platform's rules.
FAQ sources: Proxyway Market Research 2026, ScrapingBee’s test, Bright Data release notes, and the pricing pages linked in each provider section.
Sources
37 sources, accessed .
Official pricing, docs, changelogs and status pages 29
- Oxylabs mobile proxies: tiers, pool, terms
- Oxylabs general conditions (PAYG refund conditions)
- Oxylabs residential pricing (for the mobile vs residential gap)
- Decodo mobile pricing and FAQ
- Decodo residential pricing, trial and money-back terms
- Decodo licence agreement (credit expiry)
- SOAX pricing: geo tiers, plans, expiry, refunds
- SOAX mobile proxies
- SOAX residential proxies (pool claim)
- Evomi mobile proxies: plans, targeting, sessions
- Evomi pricing (PAYG expiry, residential Core)
- Evomi refund policy
- Byteful mobile proxies: tiers, pool, sourcing
- Byteful residential pricing (for the mobile vs residential gap)
- DataImpulse mobile proxies: plans, pool, targeting
- DataImpulse residential pricing (for the mobile vs residential gap)
- DataImpulse terms (refund scope)
- IPRoyal mobile proxies: tiers, dedicated ports, carriers
- IPRoyal terms of service v10
- NodeMaven pricing (residential and mobile plan)
- Proxidize pricing
- Proxidize mobile proxies
- Rayobyte pricing (mobile plans)
- Bright Data mobile URL (lands on a general proxy page)
- Bright Data mobile pricing URL (HTTP 404)
- Bright Data release notes, 2026-10-24
- netnut.io (observed page state)
- FBI public service announcement I-031226-PSA on residential proxy networks
- Source group announcement (Oxylabs, Decodo, Webshare, Honeygain, ScrapingBee)
