In August 2024 NIST published its first post-quantum cryptography standards. They have unfamiliar names and some unfamiliar properties. This field guide explains what each algorithm is for and what it costs in size and performance.
The line-up
| Standard | Algorithm | Based on | Purpose | Maths family |
|---|---|---|---|---|
| FIPS 203 | ML-KEM | CRYSTALS-Kyber | Key encapsulation (key exchange) | Module lattices |
| FIPS 204 | ML-DSA | CRYSTALS-Dilithium | Digital signatures | Module lattices |
| FIPS 205 | SLH-DSA | SPHINCS+ | Digital signatures (conservative) | Hash functions |
| FIPS 206 (in development) | FN-DSA | Falcon | Compact digital signatures | NTRU lattices |
| Selected March 2025 | HQC | HQC | Backup key encapsulation | Error-correcting codes |
ML-KEM: the new key exchange
ML-KEM is a key encapsulation mechanism (KEM). One party uses the other's public key to encapsulate a random shared secret; the holder of the private key decapsulates it. Both end up with the same secret, which then keys symmetric encryption such as AES. It replaces RSA key transport and elliptic-curve Diffie–Hellman in protocols like TLS, SSH and VPNs.
| Parameter set | Security category | Public key | Ciphertext |
|---|---|---|---|
| ML-KEM-512 | 1 (≈ AES-128) | 800 bytes | 768 bytes |
| ML-KEM-768 | 3 (≈ AES-192) | 1,184 bytes | 1,088 bytes |
| ML-KEM-1024 | 5 (≈ AES-256) | 1,568 bytes | 1,568 bytes |
For comparison, an X25519 public key is 32 bytes. ML-KEM is very fast, often faster than elliptic-curve operations, but its messages are much larger. ML-KEM-768 is the common default, for example in the hybrid X25519MLKEM768 group used in TLS. ML-KEM-1024 is required in some government profiles such as the NSA's CNSA 2.0.
ML-DSA: the general-purpose signature
ML-DSA replaces RSA and ECDSA signatures for most purposes: certificates, code signing, document signing and authentication.
| Parameter set | Security category | Public key | Signature |
|---|---|---|---|
| ML-DSA-44 | 2 | 1,312 bytes | 2,420 bytes |
| ML-DSA-65 | 3 | 1,952 bytes | 3,309 bytes |
| ML-DSA-87 | 5 | 2,592 bytes | 4,627 bytes |
An ECDSA P-256 signature is about 64 bytes, with a 64-byte public key. Signing and verifying are fast, but certificate chains and protocols that carry several signatures grow significantly. That's why TLS certificate migration is harder than key-exchange migration.
SLH-DSA: the conservative backup
SLH-DSA's security rests only on hash functions, which are very well understood. That makes it a hedge if a weakness is ever found in lattice cryptography. The trade-off is size and speed: signatures range from about 7.8 KB (the "small" 128-bit variant) to around 50 KB at the highest levels, and signing is slow.
It suits long-lived, infrequent signatures where trust must last decades, such as firmware roots of trust, but not high-volume use like TLS handshakes.
The older stateful hash-based schemes LMS and XMSS (NIST SP 800-208) are also approved and widely used for firmware signing. They require careful state management, since a one-time key must never be reused.
FN-DSA: compact signatures, coming
Falcon-based FN-DSA offers smaller signatures than ML-DSA (around 666 bytes at the lower level), which helps where size matters. Its implementation is trickier, particularly the floating-point arithmetic, which needs care to avoid side-channel leaks. The standard is still being finalised.
HQC: a different mathematical bet
NIST selected HQC in March 2025 as a backup KEM based on error-correcting codes rather than lattices. If lattice-based ML-KEM were ever weakened, HQC would be ready. A draft standard is expected first, then a final one.
Hybrid deployments
During the transition, many systems combine a classical and a post-quantum algorithm:
- Hybrid key exchange: for example X25519 + ML-KEM-768, where the shared secret derives from both. If either is secure, the session is secure.
- Composite or dual signatures: for example ECDSA + ML-DSA, under active standardisation for certificates.
Hybrid protects against both a future quantum computer and the possibility of an undiscovered flaw in the newer algorithms.
Choosing, in one paragraph
For key exchange, use ML-KEM-768 in hybrid with X25519, unless a profile requires ML-KEM-1024. For general signatures, plan for ML-DSA-65 (or ML-DSA-87 where required), probably in composite form for certificates during transition. For firmware and long-term roots of trust, consider LMS/XMSS or SLH-DSA. Above all, use well-reviewed libraries; never implement these algorithms yourself.
Learn more
See post-quantum cryptography explained and the migration roadmap. For PKI and cryptography fundamentals, try the free CISSP track.
Written and reviewed by the DigiTrust Asia editorial team. How we research, review and correct articles.
Strengthen your crypto fundamentals
Cryptography lessons and practice questions in the free CISSP track.
Create a free account