Product comparison
Auris, Signal and SimpleX
Choose the approach that fits the conversations you want to have. Here is what Auris puts first and how specific design choices compare.
An identity independent of your phone number
Auris creates an ID on your device and lets you connect without providing a phone number or email. For someone who wants phone-number-free registration, this is a concrete advantage over Signal on iPhone: Signal's released apps still require a phone number, and its phone-number-less registration is a paid Android beta (a one-time US$2.99 fee through Google Play billing, with zero-knowledge payment receipts) as of September 2026. Signal's usernames let people connect without sharing a number with each other.
SimpleX also avoids a required phone number and goes further in its core addressing model by avoiding mandatory network-wide user IDs. Auris uses a random directory ID alongside its sealed-send and per-contact queue mechanisms.
What Auris puts first
- Calls at the center
- A native iPhone experience built around voice and video, adaptive media and connection recovery. Explore calling.
- Privacy you choose
- Invitation-only discovery, relayed calls, configurable notifications and a separate history key with biometric and passphrase options. Explore the controls.
- Future-facing cryptography
- Hybrid classical and post-quantum key material, with research into authentication and recovery. Inspect the construction.
These are the priorities behind Auris. The table below names the documented mechanisms so you can compare the choices in detail. Cross-app call-quality measurements are future work, rather than a measured advantage claimed here.
Protocol and privacy details
| Dimension | Auris | Signal documentation | SimpleX documentation |
|---|---|---|---|
| First-contact authentication | Released pv4 seals a transcript signed with Ed25519 and ML-DSA-65. Human identity still needs a trusted binding; full computational composition is open. | PQXDH rev. 3 describes classical mutual authentication and post-quantum forward secrecy. Its active-quantum authentication limit is explicit. | The layered protocol reference supplies a different connection model. A matching active-quantum authentication claim is not established by this review. |
| Ongoing PQ key updates | Custom periodic and continuous mechanisms. Directional sketch and sampled simulations; no universal healing deadline. | Double Ratchet rev. 4 includes SPQR and Triple Ratchet; ML-KEM Braid has a separate analysis and traffic-dependent recovery discussion. | The current protocol-layer documentation describes post-quantum Double Ratchet encryption. Exact recovery and compromise claims require its specific construction and assumptions. |
| Network-wide user ID | A random Auris ID remains in directory and identified fallback/call paths. Anonymous queues omit owner IDs from their records. | Phone-number registration (a paid number-free registration is in Android beta since September 2026) with optional usernames for initiating contact. Signal states that, given the plaintext of a username in use, it can connect that username to the account. Number visibility and discovery have separate settings, described in Signal Support. | Network design uses queue addressing without mandatory network-wide user identifiers. Its July 2026 public-names announcement describes optional discovery names for channels, businesses and creators: plaintext labels registered on an Ethereum-derived registry (an ENS fork) and resolved through a names router that sees which name was looked up. Those are a separate layer, and SimpleX says private users do not need them. |
| Transferable initial evidence | The recipient can open a signed initial. Sealing does not establish deniability. | PQXDH describes scoped forms of deniability and distinguishes classical and quantum questions. | No same-model deniability verdict is asserted here. |
| Routing privacy | Sealed sends and queues remove explicit fields. IP, timing, wake and fallback correlation remain. | Sealed Sender describes hiding a sender field from delivery. This alone is not a global traffic-analysis claim. | Pairwise queues and routing separation are documented. Identifier minimization is distinct from resistance to all correlation. |
| Assurance | No independent product audit. Public R2 covers a separate v5 candidate; R3 and other internal models are scoped project evidence. | The published specifications link formal analyses with their assumptions. Their existence does not eliminate implementation or composition questions. | The 2024 design-review record links an independent Trail of Bits report and responses. Its scope is not a timeless whole-app guarantee. |
Which versions are being compared?
Checked September 5, 2026; the Signal registration and username-lookup cells were re-checked October 29, 2026. Auris refers to the recorded build-147 beta and explicitly identified build-148 source changes. Signal references PQXDH revision 3, Double Ratchet revision 4 and ML-KEM Braid revision 1. SimpleX references the architecture pages retrieved on that date and its dated design-review record.
This is a comparison of documented mechanisms and evidence. It does not independently attest which protocol path every installed Signal or SimpleX client currently selects. Unverified deployment details remain unverified.
How to read the evidence
protocol + version + property + attacker + compromise timing + delivery assumptions + evidence + unresolved boundary
Adding ML-DSA to a transcript changes authentication and transferable evidence, but does not automatically prove secrecy under every compromise scenario. Calling a ratchet “continuous” does not establish a faster recovery distribution. Removing an ID does not prove traffic indistinguishability. Each requires its own argument and experiment.
The Auris ledger supplies its current assurance limits; the primary-source library lets readers inspect the original statements. No cross-app call-quality or overall security ranking is established by this review.
Reviewed · Review scope and versions