TL;DR: Platforms detect virtual (VoIP) numbers with a four-layer stack that runs in under 200 milliseconds: an HLR line-type lookup, Line Type Intelligence enrichment, an MNP porting-history check, and IP-plus-device fingerprinting. A VoIP number from Google Voice, Twilio, Telnyx, Bandwidth, or Plivo returns “voip” at layer 1 and is hard-blocked before a code is sent. A real carrier-issued SIM on Vodafone, O2, T-Mobile, or Lebara returns “mobile,” carries clean tenure and porting history, and passes every layer, consistently high success. VirtualSMS runs real SIM cards (not VoIP) across 2500+ services in 145+ countries, from single-use verifications to multi-day rentals, with a 20-minute auto-refund and MCP + API access.
Between 2023 and 2025, every major consumer platform quietly upgraded its phone-verification infrastructure. Nobody announced it, you only see it in the rejection rates. SMS services that reported 70-80% pass rates in 2023 dropped to 20-40% for VoIP-pool numbers by early 2025, while real-SIM services held consistently high. The mechanism behind that gap is a four-layer detection stack that fires in under 200 milliseconds every time someone submits a phone number.
If you want the short version: platforms check what the number actually is at the network level before they send a single digit of an OTP. A virtual number can be spun up in milliseconds via an API, and that is exactly what gives it away. A number that flexible cannot also look, to an HLR query, like a physical SIM that registered on a carrier weeks ago.
Key Takeaways
- The detection stack has four independent layers, HLR line type, Line Type Intelligence, MNP porting flags, and IP/device fingerprinting, and a number can pass one while failing another.
- VoIP numbers are hard-blocked at layer 1 on WhatsApp, Telegram, Coinbase, Cash App, and gov.uk One Login: the code is never sent.
- “Non-VoIP” is not the same as “real SIM”, prepaid and MVNO pools pass the HLR check but fail Line Type Intelligence on age and reuse.
- Real carrier SIMs return
line_type: "mobile"and pass all four layers, which is why real-SIM activations run at consistently high rates versus 20-40% for VoIP on strict platforms.- The economics favor real SIM at any real scale: 100 verified accounts need ~105-110 real-SIM activations versus 250-500 VoIP attempts.
What Is the 4-Layer Detection Stack Platforms Use?
Platforms detect virtual phone numbers with four independent checks, an HLR line-type lookup, Line Type Intelligence enrichment, an MNP porting-flag query, and IP-plus-device fingerprinting, that together run in under 200 milliseconds. Each layer tests something different, and each can block a number the others let through. That independence is the whole point: it is why a number can clear the raw HLR check and still fail on carrier reputation or porting history.
- Detection stack (definition)
- The ordered set of real-time checks a platform runs on a submitted phone number before dispatching an OTP, verifying line type, carrier reputation and number age, porting history, and the coherence between the submitting device and the number’s country.
Here is the whole stack in one view, what each layer does, what it blocks, what superficially fools it, and what actually beats it.
| Detection layer | How it works | What it blocks | What superficially fools it | What beats it reliably |
|---|---|---|---|---|
| HLR line type | Queries the Home Location Register for mobile / landline / voip / prepaid by the number’s carrier (OCN) | All VoIP-range numbers, Google Voice, Skype, Twilio, Telnyx, Bandwidth, Plivo pools | VoIP-to-mobile ports (partially); some prepaid burners | A real SIM registered on a major mobile operator |
| Line Type Intelligence | Enriches HLR with carrier reputation, number-age estimate, and cross-account reuse score | Flagged carrier ranges; numbers under ~30 days old; high-reuse pool numbers | Freshly provisioned VoIP (briefly passes age); numbers ported from clean carriers | A real SIM with weeks of activity on a high-reputation carrier |
| MNP porting flag | Checks Mobile Number Portability history, original carrier and whether it was ported from a VoIP provider | VoIP numbers ported to mobile carriers; port history under ~60 days | Long-ported numbers whose timestamp ages out of the lookback window | Real SIM numbers never ported, or ported mobile-to-mobile between major carriers |
| IP + device fingerprint | Cross-references the device’s IP country, ASN type, and browser/app fingerprint against the number’s carrier country | Distant-country IP + number combos; datacenter ASN + mobile number | Same-country residential proxy or VPN (partially) | A coherent IP-country pairing matching the number’s carrier country |
Why non-VoIP numbers pass verification and VoIP numbers do not →Citation Capsule: Modern SMS verification runs a four-layer detection stack in under 200 milliseconds: (1) an HLR line-type lookup returning mobile/landline/voip/prepaid from carrier registration; (2) Line Type Intelligence enrichment via Twilio Lookup and Telesign Phone ID adding carrier reputation, number age, and reuse scoring; (3) an MNP porting-flag query detecting VoIP-to-mobile port history; (4) IP and device fingerprinting matching the submitting device’s country against the number’s carrier country. The layers are independent, a number can pass one and fail another. VoIP numbers are blocked at layer 1; real carrier-issued SIMs return “mobile” and satisfy every layer, which is why real-SIM pass rates hold consistently high while VoIP-pool rates fell to 20-40% on strict platforms through 2025.
Layer 1: How Does the HLR Line-Type Lookup Work?
The Home Location Register is a core part of the GSM and LTE network. Every subscriber number maps to an HLR record with the home carrier (its OCN, or Operating Company Number), the IMSI tied to the SIM card, and a line-type classification. When a platform runs an HLR query, the response includes a line_type field with one of: mobile, landline, voip, prepaid, or unknown.
For WhatsApp, Telegram, Coinbase, Cash App, and gov.uk One Login, a voip result is a hard block, no code is sent. For lighter platforms like Discord and Instagram, voip is a soft warning that feeds a risk score rather than an outright wall.
What counts as VoIP at the HLR level? Any number provisioned by a software-based telephony provider, Twilio, Telnyx, Plivo, Bandwidth, vapi.ai, and the consumer services built on them (Google Voice, Skype online numbers, TextNow, TextFree, OpenPhone), resolves to a VoIP OCN. The HLR is not fooled by number formatting or country-code tricks: it queries actual network registration, not the number’s structure.
Technical note: HLR queries run in real time over SS7 signaling or via REST-aggregated services (Twilio Lookup, Numverify, Numlookup). As of 2026, Twilio Lookup’s Line Type Intelligence endpoint returns line_type: "voip" for the overwhelming majority of VoIP-pool numbers within milliseconds. Platforms that integrated it in 2023-2024 saw immediate accuracy jumps, the primary driver of the pass-rate cliff for VoIP services.
Layer 2: What Does Line Type Intelligence Add on Top?
Line Type Intelligence (LTI) is the enrichment layer over raw HLR data. Providers like Twilio Lookup and Telesign Phone ID combine the HLR line type with carrier-reputation databases (Nomorobo, Hiya), a number-age estimate, and historical reuse rates across their customer networks. The result is not a yes/no line type, it is a composite risk score.
How number-age scoring works
LTI APIs estimate a number’s age from the first time it appeared in carrier signaling records. A real SIM active on a carrier for months accumulates tenure. VoIP numbers provisioned on demand and recycled fast have very short tenure, often under 24 hours between first observation and deactivation. Platforms flag tenure under 30 days as elevated risk and under 7 days as high risk.
How behavioral reuse tracking works
When a platform records a number being submitted for many account creations, it increments that number’s reuse counter. VoIP pools recycle numbers across thousands of activations, so a “fresh” VoIP number can arrive already carrying a high reuse count. Real SIM numbers allocated per activation from managed inventory carry zero prior reuse on that platform.
Common misconception: “non-VoIP” does not guarantee a real SIM. Some providers market numbers as non-VoIP while using MVNO or prepaid pools that LTI classifies as prepaid or landline. Those pass the layer-1 HLR check (not VoIP) but fail layer-2 scoring on high reuse and low tenure. The distinction that matters is real SIM on a major carrier, not merely “not VoIP.”
Citation Capsule: Line Type Intelligence APIs (Twilio Lookup, Telesign Phone ID) enrich raw HLR line-type data with three fraud signals: carrier-reputation scores from Nomorobo and Hiya, number-age estimates from first-seen signaling records, and cross-account reuse counts. VoIP-pool numbers score badly on all three, flagged carrier ranges, sub-24-hour tenure, and high reuse from recycled pools, while a real carrier SIM scores clean on each. Critically, a number marketed as “non-VoIP” can still fail LTI if it comes from a prepaid or MVNO pool with high reuse and low tenure; passing verification reliably requires a real SIM on a high-reputation mobile operator, not just the absence of a VoIP flag.
Layer 3: What Do MNP Porting Flags Detect?
Mobile Number Portability (MNP) lets subscribers keep a number when switching carriers. Porting leaves an auditable trail, originating carrier, receiving carrier, and completion timestamp, recorded in national MNP databases. Platforms that query MNP can tell whether a number was originally a VoIP number later ported into a mobile carrier, a pattern strongly correlated with fraudulent account creation.
Mobile carriers in several European markets actively block port-in requests from VoIP or software-defined carriers, so a Telnyx or Bandwidth number cannot be ported onto those mobile networks at all. The practical effect: real SIM numbers on those carriers carry clean porting histories, either no port record (originally assigned mobile) or clean mobile-to-mobile ports between legitimate operators.
For government identity platforms like gov.uk One Login and US ID.me, an MNP porting-flag query is a mandatory step: they can reach national MNP databases and flag any number with a VoIP origination record, even one since ported to mobile. Standard consumer platforms typically use a 60-90 day lookback; government-aligned ones keep longer windows.
Layer 4: How Do IP and Device Fingerprinting Work?
The fourth layer looks at the submitting device, not the number. Platforms compare the IP address and device fingerprint against the number’s claimed country and carrier. A UK number submitted from a datacenter IP in Singapore trips an immediate mismatch. A number submitted from an ASN belonging to a cloud host, rather than a residential ISP or mobile carrier, raises the risk score.
Device fingerprinting extends this to browser and app signals: user-agent consistency, WebRTC IP-leak detection, canvas fingerprint, and installed font set. Platforms like Tinder and OnlyFans combine device fingerprints with number submissions to catch users who cycle through numbers on a consistent device, the signature of account farming. Financial apps like Cash App and Coinbase apply especially strict IP-to-number country matching.
This is the one layer a residential proxy partially addresses, a residential IP in the same country resolves the geographic mismatch. But proxies do nothing for layers 1-3, which query the number independently of where the submission comes from. A VoIP number submitted from a clean residential US IP is still blocked at layer 1.
Why Do Real SIM Numbers Pass All Four Layers?
A real SIM number passes because it authentically is what each layer tests for, there is no spoofing involved. It is registered on a physical carrier network, carries an IMSI tied to an actual SIM card, maps to a legitimate mobile-operator OCN, and has a network history dating from first activation. Here is what each layer sees.
| Detection layer | What VoIP / disposable returns | What a real SIM returns | Platform decision |
|---|---|---|---|
| HLR line type | line_type: "voip" | line_type: "mobile" + real carrier OCN | Pass, qualifies for code send |
| LTI enrichment | High reuse, low tenure (under 7 days), flagged range | Low/zero reuse, weeks of tenure, clean reputation | Pass, risk stays below threshold |
| MNP porting | VoIP origination; recent port; port-in blocked | Mobile origination; no VoIP ancestry; stable assignment | Pass, no fraud-pattern port history |
| IP + device | Datacenter or mismatched-country IP | Number’s country matches a coherent submission context | Pass when submission context is coherent |
| Overall pass rate | 20-40% on strict platforms | Consistently high on real-SIM activations | , |
The consistently high pass rate is not leniency, it is authenticity. The number is exactly what it claims to be at the network level. The remaining sub-5% of real-SIM failures usually trace to platform-specific temporary blocklists, carrier-level SMS outages, or an unusual platform pairing, not the detection stack itself.
This is why real SIMs matter most for AI agents and automation that verify across many platforms: at a 20-40% VoIP pass rate, an agent has to retry across 3-5 numbers per account, and the economics collapse. Real-SIM verification via MCP or API is the only path that stays viable at production volume.
Citation Capsule: A real carrier-issued SIM passes all four detection layers because it satisfies each authentically rather than spoofing it: the HLR returns
line_type: "mobile"with a legitimate mobile-operator OCN; Line Type Intelligence sees clean carrier reputation, real tenure, and low per-platform reuse; MNP shows mobile origination with no VoIP ancestry; and IP/device fingerprinting matches when the submission context is coherent. The result is a consistently high pass rate on real-SIM activations versus 20-40% for VoIP on detection-strict platforms, a difference driven entirely by network-level authenticity, not by platforms treating real SIMs more leniently.
Which Platforms Detect Hardest in 2026?
Detection intensity varies sharply by platform type. Financial services, government identity, and real-time messaging apps run the most comprehensive stacks; social and gaming platforms generally run lighter checks. Here is the current posture by category.
| Platform | Layers active | VoIP outcome | Real SIM outcome |
|---|---|---|---|
| HLR + LTI + reuse + device | Hard block at HLR; code never sent | Pass; code arrives within ~60s | |
| Telegram | HLR + reuse | Blocked or flagged for monitoring | Pass |
| Cash App | HLR + LTI + carrier requirement + compliance | Hard block; real SIM required by policy | Pass with a real-SIM number |
| Coinbase | HLR + LTI + carrier reputation + identity cross-ref | Hard block; financial compliance | Pass; major carrier required |
| Robinhood | HLR + LTI + compliance | Hard block | Pass |
| Tinder | HLR + device fingerprint + reuse | Hard block or silent failure | Pass; real mobile-carrier numbers perform best |
| gov.uk One Login | HLR + MNP porting + national database | Hard block; government-grade validation | Pass with a real carrier SIM |
| ID.me | HLR + MNP porting + identity cross-ref | Hard block | Pass with a real carrier SIM |
| OnlyFans | HLR + reuse + device fingerprint | Blocked or reuse-flagged | Pass |
| Discord | HLR (soft) + behavioral | Soft flag; some VoIP passes briefly | Pass consistently |
| HLR + reuse | Variable; stricter in 2025-2026 | Pass | |
| Most gaming / mid-tier social | Format check only | Often passes (no HLR query) | Pass |
The pattern is consistent: any platform with fraud risk or KYC requirements integrates line-type APIs. The question is not whether a VoIP number fails, it is how many attempts it takes before you accept that it will.
Browse 2500+ services and live availability →What Do VoIP Providers Actually Offer, and Why Does It Fail Detection?
Several providers market “virtual phone number” services for verification. Understanding what they genuinely provide clarifies why they fail strict platforms.
- Twilio: Programmable SMS numbers (long/short codes from the console) resolve to VoIP line types in HLR queries, including via Twilio’s own Lookup API. Using one for account verification means HLR rejection anywhere a Lookup runs.
- Telnyx: VoIP cloud numbers that resolve to
line_type: "voip". Even its “real carrier” partnerships originate from software-provisioned OCN ranges, detectable via MNP origination records. - Bandwidth: A carrier-grade VoIP provider whose OCNs reputation databases have extensively catalogued as VoIP/CLEC ranges. Numbers on a Bandwidth OCN fail carrier-reputation scoring.
- Plivo: Long code, short code, and toll-free numbers over VoIP infrastructure, resolving to VoIP line types and flagged OCNs. Built for outbound notifications, not verification receipt.
- vapi.ai: Voice-AI infrastructure on VoIP telephony; numbers come from VoIP pools and fail layer 1 on all strict platforms.
The core issue: these providers are in the business of provisioning telephony for communication apps, and their products are correctly designed for that. The failure mode is using communication infrastructure for an identity use case. Flexible software provisioning and carrier-grade identity signals are fundamentally in tension, a number that can be spun up in milliseconds cannot also look, to an HLR, like a physical SIM registered weeks ago.
VoIP vs Real SIM: When Is Each Appropriate?
Not every verification needs a real SIM. It depends on whether the target platform runs HLR/LTI detection and what a failed verification costs you.
| Use case | Platform detection | VoIP viable? | Real SIM required? |
|---|---|---|---|
| Internal dev / QA of your own SMS flow | None (your platform) | Yes, full control | No |
| Low-stakes apps (gaming, mid-tier social) | Format check only | Often yes | No, but real SIM has higher success |
| WhatsApp, Telegram, Discord, Instagram | HLR + behavioral | No, blocked at HLR | Yes |
| Fintech: Cash App, Coinbase, Robinhood | Full 4-layer + compliance | No, hard block | Yes |
| Government identity: gov.uk One Login, ID.me | HLR + MNP + national database | No, hard block | Yes |
| AI agent automation across platforms | Mixed | No, economics break at 20-40% | Yes, via MCP or API |
The economic argument at scale is decisive: at a consistently high real-SIM pass rate versus 20-40% for VoIP, 100 successful verifications need roughly 105-110 real-SIM activations versus 250-500 VoIP attempts. The apparent per-number price advantage of VoIP reverses once you count retries, and it does not count the platform bans that pile up from repeated failed VoIP attempts.
Which VirtualSMS Option Fits Your Verification?
Every VirtualSMS number is a real carrier-issued SIM, not VoIP, so all three options below return line_type: "mobile" and pass the detection stack described above. The right one depends on how long you need the number.
- Single verification: one OTP code across 2500+ services in 145+ countries. Auto-refund if no SMS arrives within 20 minutes. Right for a one-time account creation or a single KYC step.
- Full Access Rental: an entire local SIM, usable for any service, rented exclusively to you for 1, 3, 7, 14, or 30 days. Every SMS routes to your private inbox. Right for developers integrating SMS workflows or teams testing across many services at once.
- Platform Rental: a partner/global number locked to one service for 1, 3, or 7 days, with a 20-minute refund window if no SMS arrives. Cheaper because you pay for one service slot, not the whole SIM. Right for ongoing management of a single account over days.
All three run on real SIM cards on operators like Vodafone, O2, T-Mobile, and Lebara, not VoIP, and integrate via MCP or API for automated and agent workflows.
Compare Full Access and Platform Rental → See full pricing →Frequently Asked Questions
How do platforms detect if a phone number is VoIP or virtual?
Platforms run a real-time detection stack, usually four layers deep, in under 200 milliseconds. Layer 1 is an HLR (Home Location Register) lookup that returns the number’s line type, mobile, landline, voip, or prepaid, straight from carrier network registration. Layer 2 is Line Type Intelligence: APIs like Twilio Lookup and Telesign Phone ID enrich the HLR result with carrier reputation, number age, and cross-account reuse scoring. Layer 3 checks Mobile Number Portability history for a VoIP origination record. Layer 4 fingerprints the submitting device’s IP country and ASN against the number’s carrier country. A VoIP number returns “voip” at layer 1 and is blocked before an OTP is ever dispatched. A real carrier SIM returns “mobile” and passes.
What is an HLR lookup and how do verification platforms use it?
HLR stands for Home Location Register, the carrier network database that maps a phone number to its operator, IMSI, and line type. When WhatsApp, Coinbase, or Cash App calls a lookup API (Twilio Lookup, Telesign, Numverify), the structured response includes a line_type field: mobile, landline, voip, or prepaid. A line_type of “voip” is an immediate hard block on strict platforms, no code is sent. The HLR queries actual network registration, not number formatting, so country-code tricks and reformatting do not change the result. Only a number genuinely registered on a carrier network returns “mobile.”
Can a VoIP or virtual number still pass SMS verification in 2026?
Rarely, and far less reliably than before. Low-stakes platforms that run only a format check (country code plus length) still accept VoIP numbers. But every high-value platform, WhatsApp, Telegram, Cash App, Coinbase, Robinhood, Tinder, gov.uk One Login, ID.me, runs HLR or Line Type Intelligence queries and blocks VoIP at layer 1. Reported VoIP pass rates on detection-strict platforms fell to roughly 20-40% by 2025. For anything involving money, identity, or real-time messaging, a VoIP number fails consistently and a real carrier SIM is required.
What makes a real SIM number different from a virtual number?
A real SIM number is registered on a physical carrier’s network. It carries an IMSI tied to an actual SIM card and an OCN that maps to a legitimate mobile operator, so HLR queries return line_type: "mobile" with a real carrier name. A virtual (VoIP) number is provisioned in software, no physical SIM, no carrier network registration, so its HLR response returns “voip” or “landline” and resolves to a VoIP-range OCN that reputation databases have flagged. That single network-level difference is why real SIMs pass all four detection layers and VoIP numbers fail at the first.
Why does my verification code never arrive even with a “non-VoIP” number?
“Non-VoIP” on a label does not guarantee a real SIM. Some services market numbers as non-VoIP while using prepaid or MVNO pools that pass the HLR check but fail Line Type Intelligence on two other signals: number age (activations under 30 days score as elevated risk) and reuse (pooled numbers arrive already used across many accounts). A third cause is an IP-country mismatch, submitting a UK number from a distant datacenter IP trips the device-fingerprint layer. Real SIMs allocated per activation from managed inventory, submitted from a coherent location, avoid all three.
Does the country of the phone number affect verification success?
Yes. Fraud-detection rules are weighted by country. Numbers on operators with strong carrier reputations and clean portability rules, a German number on Telekom or Vodafone, a UK number on O2 or Vodafone, carry high pass rates. Numbers from markets with VoIP-heavy telco ecosystems or lax portability score lower. For critical verifications (Coinbase, gov.uk One Login, financial KYC), choosing a number from a high-reputation carrier market materially improves the pass rate. VirtualSMS lets you pick the country and see live availability across 145+ countries before you commit.
When should I use a rental instead of a single verification?
Use a single verification when you need one OTP for a one-time account creation or KYC step, from real carrier SIMs across 2500+ services, with an auto-refund if no SMS arrives in 20 minutes. Use a rental when you need the same number to keep receiving SMS over time. VirtualSMS offers two rental tiers: Full Access, an entire local SIM, any service, for 1, 3, 7, 14, or 30 days, and Platform Rental, a partner/global number locked to one service for 1, 3, or 7 days with a 20-minute refund window. Both run on real carrier SIMs and pass the same detection stack a single verification does.
The Bottom Line
Platforms detect virtual phone numbers with a four-layer stack, HLR line type, Line Type Intelligence, MNP porting flags, and IP/device fingerprinting, that fires in under 200 milliseconds and blocks VoIP at the very first layer. That is not a policy preference; it is what carrier-grade lookup APIs return. VoIP numbers fail because they are provisioned in software and cannot present the network-level signals a real SIM presents.
Real SIM numbers pass because they authentically are what each layer tests for: a line_type: "mobile" result, clean carrier reputation, real tenure, no VoIP porting ancestry. That is the whole reason real-SIM activations hold consistently high while VoIP has slid to 20-40% on strict platforms, and why, at scale, real SIM is cheaper per verified account despite the higher per-number price.
VirtualSMS runs real carrier-issued SIM cards, not VoIP, across 2500+ services in 145+ countries, from single verifications to Full Access and Platform rentals, with MCP and API access for automated workflows. Most activations complete in under 60 seconds, and if yours does not arrive within 20 minutes, the refund is automatic. Related reading: why non-VoIP numbers pass verification and MCP vs API for SMS verification.
