Spojiti vs WhatsApp, Telegram, Signal, Session, Threema and SimpleX

Against the apps everyone already uses

Where it counts	WhatsApp	Telegram	Signal	Spojiti
IDENTITY
Works without a phone number	No	No	Partial beta, Android, paid	Yes
No email, number or sign-up form	No	No	No	Yes
Never uploads your contact book	No	No	Partial	Yes
CONTENT & CRYPTOGRAPHY
End-to-end encrypted by default	Yes	No opt-in only	Yes	Yes
Forward secrecy: a key stolen later cannot read past traffic	Yes	Partial	Yes	Yes
Encrypted group messaging	Yes	No	Yes	Yes
Call setup itself is encrypted	No	No	Yes	Yes
WHAT THE OPERATOR KNOWS
Cannot build your social graph^†	No	No	Partial	Partial
Keeps no record of any group	No	No	No encrypted group state	Yes but see †
Cannot tell a call is happening^†	No	No	Partial	Partial
MONEY
Paying is not linked to your account	N/A nothing to buy	No	N/A free; donations	Yes
No ads in the app today	No	No	Yes	Yes

^† Marked as of 11 October 2026. Spojiti seals the sender: on the sealed lane the relay receives an envelope addressed to a recipient's random routing id, and the sender's identity travels inside the ciphertext. The social-graph row is still Partial, for us and for Signal, for the reasons our White Paper (section 7) gives: an envelope to an app that cannot yet open sealed envelopes carries its sender until delivered, connection addresses and timing can be correlated, and a group message's batch shows the relay which accounts receive copies together. No group record is kept: nothing the relay stores names a group or its members. The call row is Partial at best: a push that wakes a killed app for a call has to be a high-priority push, and that alone reveals that call signalling is in flight, no content, no identity. Every app that rings a locked phone makes the same trade, and we mark ourselves the same way.

Against the privacy-native field

These are the projects a security researcher actually compares us to, row by row and sourced. Where a project does something better than we do today, the row says so, and the last card on this page says what ships to close it.

The serious comparison	Session	Threema	SimpleX	Spojiti
IDENTITY
No phone number or email required	Yes	Yes	Yes	Yes
No persistent identifier of any kind	No	No	Yes the most radical design in the field	No
CRYPTOGRAPHY
Forward secrecy today	No V2 still in design	Yes	Yes	Yes
Post-quantum encryption shipped	No	Partial	Yes	Yes hybrid ML-KEM-1024
Post-quantum ratchet inside the session	No	No	Yes	Yes ML-KEM-768, every reply
Post-quantum signatures on the account's prekeys	No	No	No	Yes ML-DSA-65 + Ed25519
OPENNESS
Fully open source	Yes	Yes	Yes	Core Spojiti Private Messenger Core · AGPL v3
USABLE AS A DAILY DRIVER
Usable free, without paying	Yes	No must buy to use	Yes	Yes
Voice & video calls	Beta shares your IP	Yes	Yes	Yes the whole app is in open beta
CAN IT PAY FOR ITSELF
Built to be funded by customers, not donations, tokens or behavioural ads	No token	Yes	No free / investor	Yes
Paying cannot be linked to your account	No	Partial store purchase	N/A nothing to buy yet	Yes blind-signed

Post-quantum, verified rather than asserted. Several projects claim post-quantum encryption. Spojiti's ML-KEM-1024, ML-KEM-768 and ML-DSA-65 implementations are checked against NIST's own ACVP known-answer vectors on your device, at every launch, before it is allowed near a conversation, and if that check fails, post-quantum switches itself off rather than pretending. The classical encryption is extended, never replaced, so an attacker has to break both.

Compiled from each project's public documentation and verified on 7 September 2026. Where any of them improves, we update this page. Sources: Session · Threema · SimpleX. Think a cell is wrong? Tell us and we will correct it.

All seven, feature by feature

The two tables above pick the rows where we do well. This one does not pick. It is every axis we could source across all seven apps, including the rows where another project is ahead of us today. A comparison you can only win is a comparison nobody believes.

Feature by feature, 11 October 2026	WhatsApp	Telegram	Signal	Session	Threema	SimpleX	Spojiti
IDENTITY, what you must hand over to start
Works without a phone number¹	No	Partial	Partial	Yes	Yes	Yes	Yes
No account to create at all²	No	No	No	Yes	Yes	Yes	Yes
Never uploads your contact book³	No	No	Partial	Yes	Yes	Yes	Yes
No persistent identifier held by the operator⁴	No	No	No	No	No	Partial	No
ENCRYPTION, what is actually protected
End-to-end encrypted by default, 1:1⁵	Yes	No	Yes	Yes	Yes	Yes	Yes
End-to-end encrypted group messaging⁶	Yes	No	Yes	Partial	Yes	Partial	Yes
Forward secrecy, a stolen key cannot open old messages⁷	Yes	Partial	Yes	No	Yes	Yes	Yes
Post-quantum encryption shipped⁸	No	No	Yes	No	No	Yes	Yes
Post-quantum ratchet inside the session⁸	No	No	Yes	No	No	Yes	Yes
Post-quantum signatures on the account's prekeys⁸	No	No	No	No	No	No	Yes
Encrypted voice and video calls⁹	Yes	Yes	Yes	Partial	Yes	Yes	Yes
Call SETUP is encrypted too, not just the audio¹⁰	No	No	Yes	Partial	Yes	Yes	Yes
METADATA, what the operator can still see
Operator cannot see who talks to whom¹¹	No	No	Partial	Partial	No	Yes	Partial
Operator keeps no record of groups¹²	No	No	No	Partial	Yes	Yes	Yes
Operator cannot see your IP address¹³	No	No	No	Partial	No	Yes	No
Operator cannot tell a call is happening¹⁴	No	No	Partial	Partial	Partial	Partial	Partial
Content-free push notifications¹⁵	No	No	Yes	Yes	Yes	Yes	Yes
WHAT THE APP CAN ACTUALLY DO
Disappearing messages¹⁶	Yes	Yes	Yes	Yes	Partial	Yes	Yes
View-once messages	Yes	Yes	Yes	No	No	No	Yes
Stories / status	Yes	Yes	Yes	No	No	No	Yes
Peer-to-peer payments inside a chat¹⁷	Partial	Partial	Partial	No	No	No	No switched off for now
Multi-device¹⁸	Yes	Yes	Yes	Partial	Partial	Partial	Yes
Sender-governed screenshot control¹⁹	Partial	Partial	Partial	No	Partial	No	Yes
On-device abuse guard, nothing sent to a server²⁰	No	No	No	No	No	No	Yes
THE BUSINESS BEHIND IT
Fully open source²¹	No	Partial	Yes	Yes	Partial	Yes	Core
Published independent security audit²²	Partial	Partial	Partial	Yes	Yes	Yes	In progress expected by end of 2026; the report will be published in full
Free to use²³	Yes	Yes	Yes	Yes	No	Yes	Yes
No advertising, ever²⁴	No	No	Yes	Yes	Yes	Yes	Yes
Paying for it cannot be linked to your account²⁵	N/A nothing to buy	No	N/A free; donations	No	Partial	N/A nothing to buy yet	Yes
Built to be paid for by its users, not behavioural ads, donations or a token²⁶	No	No	No	No	Yes	No	Yes

means the vendor does not offer it, or does not document it. Every mark below is sourced to the vendor's own documentation where one exists; where their marketing and their whitepaper disagree, we followed the whitepaper.

Sources and footnotes, line by line ↓

The footnotes are the honest part

Works without a phone number. Telegram always needs a number; it can be an anonymous one bought with crypto via Fragment. WhatsApp usernames (2026) hide your number from other users, registration still requires it. Signal usernames do the same. Signal began a sign-up without a number (Signal Login: an account ID, a recovery key and a one-time fee of about US$3) in its Android beta in September 2026; the standard sign-up still asks for a number, so it is marked Partial until that ships to everyone.
No account to create at all. For WhatsApp, Telegram and Signal the phone number IS the account (Signal's numberless sign-up, in its Android beta, still creates an account, with an ID and a recovery key).
Never uploads your contact book. Signal hashes contacts into an SGX enclave rather than uploading them in the clear. Threema's sync is optional and hashed. Spojiti's optional People I know reads the numbers on your phone and matches blinded fingerprints there; the contact book never leaves the phone.
No persistent identifier held by the operator. SimpleX is the only one that comes close: routing uses pairwise per-queue identifiers, so no single id spans your contacts. Session has an Account ID, Threema a Threema ID, and Spojiti a random relay routing id that names no one.
End-to-end encrypted by default, 1:1. Telegram's default cloud chats are readable by Telegram; Secret Chats are opt-in and mobile-only.
End-to-end encrypted group messaging. Telegram has no end-to-end encrypted groups at all. Session's E2EE groups cap at 100 and its Communities are not E2EE; SimpleX groups are E2EE but its Channels are visible to relay operators.
Forward secrecy, a stolen key cannot open old messages. Telegram only in Secret Chats. Session dropped forward secrecy when it left the Signal Protocol; Protocol V2 will restore it but had not shipped as of August 2026.
Post-quantum encryption shipped, ratchet and signatures. Signal ships a Kyber-1024 hybrid key agreement (PQXDH, 2023) and an ML-KEM-768 ratchet inside the session (SPQR, from October 2025, phased rollout). SimpleX ships sntrup761 at every ratchet step. Threema's February 2026 announcement is a research collaboration with IBM, its own June 2026 cryptography whitepaper contains no post-quantum content. Session's is specified for V2, not shipped. Spojiti ships ML-KEM-1024 hybrid key agreement, an ML-KEM-768 ratchet that turns with every reply, and ML-DSA-65 + Ed25519 signatures on the account's signed prekeys, post-quantum authentication of keys, which none of the other six ship today. All three are verified against NIST test vectors on the device at every launch.
Encrypted voice and video calls. Session's calls are plain peer-to-peer WebRTC and are not onion-routed, so they reveal your IP to the person you are calling.
Call SETUP is encrypted too, not just the audio. WhatsApp and Telegram route call signalling through their own servers, which learn who called whom and when, even though the call itself is end-to-end encrypted. Signal and Threema carry the SDP inside E2E messages, as we do. We previously marked Signal 'Partial' here, that was wrong and is corrected.
Operator cannot see who talks to whom. Spojiti seals the sender: on the sealed lane an envelope carries only the recipient's random routing id on the outside, and the sender's identity is inside a layer only the recipient's keys open (X25519 + ML-KEM-1024), so the relay stores no sender for it. That is Partial, not Yes, for the reasons our White Paper (section 7) gives: an envelope to an app that cannot yet open sealed envelopes carries the sender until it is delivered; the network address and timing of connections can be correlated until the private route ships; and a group message goes out as one batch of sealed copies, so the relay can see which accounts receive copies together. Signal's sealed sender has the same kind of limits and is marked the same way. SimpleX has unidirectional queues on separate servers.
Operator keeps no record of groups. Our relay has no group table: nothing it stores names a group or its members. It does carry each group message as one batch of individually sealed copies, so it can see which accounts receive copies together (White Paper, section 2.8). Signal keeps each group's state on its servers, encrypted so the server cannot read the membership; that is still a record, so it is marked No.
Operator cannot see your IP address. SimpleX routes privately by default; Session onion-routes everything except calls. Signal, Threema and Spojiti all terminate your connection directly.
Operator cannot tell a call is happening. A push that wakes a killed app for a call must be a high-priority push, so our envelopes carry a coarse wake class ('call', 'message', 'none'). It reveals no content and no identity, but it does reveal that call signalling is in flight. Every app that rings a locked phone makes this trade.
Content-free push notifications. The push payload carries no sender and no preview.
Disappearing messages. Threema offers local auto-delete on your own device rather than a negotiated two-sided timer.
Peer-to-peer payments inside a chat. WhatsApp Pay is available in only two markets. Telegram Stars buy digital goods, not person-to-person money. Signal's MobileCoin is region-limited. Spojiti's payment requests are switched off: Google Pay and PhonePe refuse a person-to-person payment that another app hands them, so payments are being designed as their own feature rather than patched.
Multi-device. Threema's phone-free desktop is still beta and its multi-device protocol ships without forward secrecy. Session syncs only about 14 days of history. SimpleX's own FAQ calls multi-device 'a complex and unsolved problem'.
Sender-governed screenshot control. WhatsApp blocks screenshots of view-once media; Telegram notifies in Secret Chats; Signal and Threema on Android offer a manual per-device toggle, and Threema Work/OnPrem administrators can enforce it. Session and SimpleX offer no sender-governed control. Ours is per-contact and sender-governed, with a master switch.
On-device abuse guard, nothing sent to a server. Content is classified on your device and the verdict never leaves it. We have not found this in any other messenger.
Fully open source. Telegram and Threema publish clients but never their server. Spojiti publishes Spojiti Private Messenger Core at github.com/Spojiti: the free text-messaging path, app and relay, everything needed to verify the zero-PII architecture and the cryptography line by line, under the AGPL v3, so anyone running a modified relay must publish the changes. Only Spojiti Private Messenger Core is published: it is what proves the design, and nothing beyond it is claimed.
Published independent security audit. Session: Quarkslab 2020, published 2021. Threema: Cure53, most recently January 2024. SimpleX: Trail of Bits 2022 and 2024. WhatsApp's public NCC Group reports cover components, not the messaging protocol; Signal publishes formal-verification work rather than full client pentests. Spojiti: An independent security audit of the full product is in progress, expected by end of 2026 and on track; the report will be published in full. It covers the full product and will be published here in full, findings included.
Free to use. Threema has no free tier, you buy the app to use it.
No advertising, ever. WhatsApp began showing ads in the Updates tab in 2025; Telegram runs sponsored messages in public channels.
Paying for it cannot be linked to your account. Among the apps you pay for, ours alone: a blind signature means we sign your access code without ever seeing it, so after you redeem it we cannot connect your account to your payment. Threema gets close by letting you buy anonymously with cash or Bitcoin. N/A where there is nothing to buy: WhatsApp and SimpleX are free, and Signal is free and takes donations.
Built to be paid for by its users, not behavioural ads, donations or a token. Signal runs on donations, Session on a crypto token, SimpleX on venture capital. Threema is paid for by the people who use it, and Spojiti is built to be, which is the only model where you are the customer rather than the inventory.

Three things on this page are ours alone

Not "we do it better", we could not find them in any of the other six.

Paying us cannot be linked to your account. Your access code is blind-signed: we put our signature on it without ever seeing it. After you redeem it, nothing we keep connects your account to your payment: the link is not withheld, it was never recorded. Threema gets closest, by accepting cash and Bitcoin for the app itself. Everyone else runs on donations, a token, venture capital, or advertising.
Money moves without us in the middle. Payment requests in a Spojiti chat are switched off for now. When they return, your own bank app fulfils them, and we never touch the money, hold it, or see it. WhatsApp Pay exists in two countries; Telegram Stars buy stickers, not groceries.
Abuse is judged on your phone. The on-device guard classifies content locally and the verdict never leaves the device, safety that costs no surveillance. Every other approach we know of sends something somewhere.

Still to come, the last steps to being indisputably the best.

The audit report, in full. An independent security audit of the full product is in progress, expected by end of 2026 and on track; the report will be published in full. It covers the full product, not only the published Spojiti Private Messenger Core, and the report lands on this site the day it is complete, findings included.
A private route for your IP. Sealed sender is shipped: the relay learns which random routing id an envelope is for and never who wrote it. What it can still see is the network address a connection came from. An optional second hop through our own second box, so the relay never sees the address you came from, is the next step.
Pairwise routing ids. Your routing id is your random account id, and it does not rotate yet; SimpleX's per-contact queues are stronger here, and rotating per-contact ids are on the protocol roadmap.
Everyone you need to reach. The rest is you: move one conversation, then the next.

We publish this list because you will find it anyway, and because a comparison page that admits nothing tells you nothing. Every line is a thing to knock off, and we would rather you watch us do it than take our word for it.

What that means in practice

Nothing to leak

We never stored a name, address book, readable message or phone number. A breach yields random ids, ciphertext and device push tokens, and for people who chose to be findable, fingerprints an attacker holding our signing key could test numbers against.

Nothing to hand over

When a lawful request arrives we comply with what we hold: no content and no contact list; random ids, keys, push tokens and connection times.

Nothing to sell

There is no behavioural advertising here and no personal data to feed it. Every feature is free; a pack that raises a limit is a small amount once a year, and that is the entire relationship.

The long version: each of them, honestly ↓

The closest competitor: Session

Session is the one other mainstream messenger that refuses to ask for your phone number, and they got there before we did. They also do two things we do not: their code is fully open source, and they run on a decentralised network with onion routing rather than company infrastructure, so no single node learns both where a message came from and where it went. If your threat model is a company that can be compelled, Session answers it by having almost no company at all. Credit where it is due.

The honest gap runs the other way on cryptography. Session does not currently have forward secrecy, by their own documentation every message is encrypted under one long-term key that never rotates unless you abandon the account, so a device compromised today can expose messages sent long ago. Their Protocol V2 adds forward secrecy and post-quantum protection, and as of late 2025 it was still in design. Spojiti ships the Double Ratchet today: keys move forward with every message, so past conversations stay sealed even if a device is later taken, and a compromise heals itself as the ratchet turns. Their documentation is equally candid that voice and video calls are beta and share your IP address with your call partner and with a foundation-operated relay server.

So the two of us made opposite bets. Session removed the company to protect metadata, and accepted weaker message cryptography and beta calls to do it. We kept infrastructure we run and maintain ourselves, which is what lets our calls, groups and media be built for everyday use with the strongest ratchet available, and removed identity instead, then built the business on subscriptions that cannot be traced back to the account that redeemed them. Both are honest answers. Ours is the one you can hand to someone who has to use it every day.

What we hold ourselves to in return: Spojiti Private Messenger Core is published under the AGPL v3 at github.com/Spojiti, and an independent security audit of the full product is in progress, expected by end of 2026 and on track. The report is published here in full when it lands, findings included, and this page is updated as the picture changes.

The honest comparison: Signal

Signal is excellent. Its cryptography is the reference the whole industry copies, ours included, openly and gratefully. We are not here to run it down.

The difference sits underneath the cryptography. Signal's standard sign-up still binds your account to a phone number, a permanent, government-linked identifier (a numberless sign-up has been in its Android beta since September 2026), and it is funded by donations. Spojiti has no personal identifier to bind, because your account is a key your own device generates, and it is built to be funded by subscriptions that cannot be traced to the account that redeemed them.

Put simply: Signal was to SMS what Spojiti is to Signal, the next step, in the same direction.

The existence proof: Threema

Threema has quietly done the thing everyone said was impossible: a Swiss messenger with no phone number, real forward secrecy, open source with reproducible builds, and a decade of profitability, funded entirely by a one-time purchase. If you want proof that people will pay for privacy rather than be paid for, Threema is it, and we admire them for it.

Two differences. Threema has no free tier, you buy before you can send a single message, which is a hard wall in any market where the person you need to reach has not heard of you. And if you buy through an app store, that purchase is tied to your store account, so the payment re-attaches an identity the app carefully avoided; its own shop, which takes cash and Bitcoin, is the way around that. Spojiti stays free for one-to-one messaging forever, prices a paid year below a couple of coffees, and severs the payment from the account by mathematics rather than policy.

The most radical of us all: SimpleX

SimpleX goes further on identity than we do, and we will not pretend otherwise. We give you an account identifier that is random and unlinked to your person; SimpleX gives you no identifier at all, each contact reaches you through its own one-off address. They are fully open source with published audits; Spojiti Private Messenger Core is published under the AGPL v3, and our independent audit is in progress, expected by end of 2026, on track. On post-quantum protection we go further, ML-KEM-1024 key agreement, an ML-KEM-768 ratchet at every reply and ML-DSA-65 signatures on the account's signed prekeys, verified on the device. On per-contact identifiers they are still ahead, and that is on our protocol roadmap.

The cost is everything around it. There is no way to look anyone up, backups are manual, there is no account recovery, groups are still maturing, and the project is free, funded by donations and investors rather than by the people using it. That is the same fragility that keeps every donation-funded messenger one bad year away from trouble. Spojiti aims at the harder target: privacy a person will actually keep using, on a business that pays for itself.

The pattern, stated against ourselves

Read those tables honestly and one shape appears. The mainstream apps keep your identity, and the ad-funded ones monetise it. The privacy-native projects give up identity but are each missing a leg: Session has no forward secrecy yet, Threema has no free tier, SimpleX has no revenue model. And every single one of them publishes its apps as open source, as we publish Spojiti Private Messenger Core, under the AGPL v3.

So here is our claim, narrowly: we are the only one attempting mainstream-grade usability, no personal identity, strongest-available cryptography, and a business funded by customers who cannot be identified, all four at once. And here is what backs it: Spojiti Private Messenger Core is published, the audit is in progress, expected by end of 2026, on track, and will be published in full, and the design this site describes can be checked in that published code, which trails the full product. Read that here, then read the code.

Move one conversation

You do not have to leave anything behind today. Pick the conversation that matters most, the one you would not want read, and start there.

Download Spojiti →

Everyone encrypts the message.Nobody else erases the person.