QNSI

Algorithm Catalog

Every PQC algorithm QNSI ships, with the receipts

14 post-quantum cryptography families across 5 mathematical foundations. Each family page details FIPS standard, security levels, key + ciphertext + signature sizes, NIST ACVP coverage, QNSI crypto-policy tier support, and use-case fit - sourced from the canonical algorithm registry the cloud KMS reads from at runtime, not from marketing copy.

Live evidence
Pair every family page with /verify/conformance for live NIST ACVP test-vector results and /benchmarks for reproducible per-algorithm performance numbers.
Cryptographic foundations
QNSI intentionally ships algorithms from multiple cryptographic foundations (lattice, code, hash, multivariate) so that a future cryptanalytic advance against any one family does not break the platform - see /security/compliance for dual-provider cross-verification policy.

NIST security levels

Levels 1-5, mapped to the tier that enforces them

NIST security category belongs to the parameter set, not the algorithm family. The policy engine rejects any parameter set below a tier's floor, so this mapping is enforced at key creation and at the edge - not aspirational copy.

LevelClassical anchorKEM (FIPS 203)Signatures (FIPS 204 / 205)Tier floor that admits it
1AES-128 key searchML-KEM-512SLH-DSA-128f / 128sdefault
2SHA-256 collisionnone - ML-KEM starts at level 1/3ML-DSA-44 (no SLH-DSA set at level 2)default
3AES-192 key searchML-KEM-768ML-DSA-65, SLH-DSA-192f / 192sdefault, strict and above
4SHA-384 collisionnonenone by standards designhash floor only (SHA-384 / SHA3-384)
5AES-256 key searchML-KEM-1024ML-DSA-87, SLH-DSA-256f / 256sall four tiers

Level 4 is a gap in the standards, not a gap in the product: NIST defined no ML-KEM, ML-DSA or SLH-DSA parameter set at category 4, so nothing here claims it. Where category-4 integrity is required, the enforced control is a hash floor (SHA-384 / SHA3-384). Level floors are enforced in the same policy engine that rejects non-final algorithms at finalized-only tiers.

Decision helper

Which level does your data need?

One question about your data's lifetime maps onto the enforced tier floors. The recommendation is a starting point for a policy conversation, not a compliance assessment.

Which NIST level does your data need?

One question, mapped straight onto the policy engine’s enforced tier floors.

Recommended crypto-policy tier
default · level 1+ floor

PQC is required for every tenant on every tier - the default floor simply admits the widest algorithm set (including level-1 parameter sets such as ML-KEM-512).

A starting point, not an assessment of your compliance obligations. The floors are the ones the policy engine enforces - see /platform#tier-ladder and /pricing.

Recommended crypto-policy tier
strict · level 3+ floor

Data that must outlive the classical era gets a level-3 floor: ML-KEM-768/1024, ML-DSA-65/87, and the 192- and 256-bit SLH-DSA sets. No classical fallback.

A starting point, not an assessment of your compliance obligations. The floors are the ones the policy engine enforces - see /platform#tier-ladder and /pricing.

Recommended crypto-policy tier
maximum · level 5+ floor

Long-lived secrets get the level-5 floor: ML-KEM-1024, ML-DSA-87, 256-bit SLH-DSA only - with HSM custody required for root keys.

A starting point, not an assessment of your compliance obligations. The floors are the ones the policy engine enforces - see /platform#tier-ladder and /pricing.

Recommended crypto-policy tier
government · level 5+ floor

The government tier enforces level 5 with FIPS-finalized algorithms only (FN-DSA excluded while FIPS 206 is a draft), requires qualified HSM custody, and cross-verifies providers.

A starting point, not an assessment of your compliance obligations. The floors are the ones the policy engine enforces - see /platform#tier-ladder and /pricing.

Select a data lifetime above to see the matching enforced tier.

Wire sizes, drawn to scale

The post-quantum size tax, visualized.

Post-quantum keys and signatures are bigger than what they replace - and the honest catalog shows the extremes, not just the flattering rows. ML-KEM-768 fits in 1,184 bytes while Classic McEliece trades 1.3 MB of public key for maximum conservatism.

ML-KEM-512 · ML-KEM public key800 B
ML-KEM-768 · ML-KEM public key1.2k B
ML-KEM-1024 · ML-KEM public key1.6k B
ML-DSA-44 · ML-DSA public key1.3k B
ML-DSA-65 · ML-DSA public key2.0k B
ML-DSA-87 · ML-DSA public key2.6k B
SLH-DSA-SHA2-128s · SLH-DSA public key32 B
SLH-DSA-SHA2-128f · SLH-DSA public key32 B
SLH-DSA-SHA2-256s · SLH-DSA public key64 B
SLH-DSA-SHA2-256f · SLH-DSA public key64 B
SLH-DSA-SHAKE-128s / 128f / 192s / 192f / 256s / 256f · SLH-DSA public key32 B
Falcon-512 · FN-DSA public key897 B
Falcon-1024 · FN-DSA public key1.8k B
HQC-128 / HQC-1 · HQC public key2.2k B
HQC-192 / HQC-3 · HQC public key4.5k B
HQC-256 / HQC-5 · HQC public key7.2k B
BIKE-L1 · BIKE public key1.5k B
BIKE-L3 · BIKE public key3.1k B
BIKE-L5 · BIKE public key5.1k B
mceliece348864 / 460896 / 6688128 / 6960119 / 8192128 · Classic McEliece public key261k B
FrodoKEM-640-AES / SHAKE · FrodoKEM public key9.6k B
FrodoKEM-976-AES / SHAKE · FrodoKEM public key16k B
FrodoKEM-1344-AES / SHAKE · FrodoKEM public key22k B
ntru_hps2048509 / 677 / 821 / 1229 · NTRU public key699 B
ntru_hrss701 / 1373 · NTRU public key1.1k B
sntrup761 · NTRU Prime public key1.2k B
MAYO-1 / MAYO-2 / MAYO-3 / MAYO-5 · MAYO public key1.2k B
CROSS-RSDP / CROSS-RSDPG (multiple param sets) · CROSS public key77 B
OV-Is / Ip / III / V (multiple param sets) · UOV public key412k B
SNOVA (multiple param sets at NIST levels 1, 3, 5) · SNOVA public key2.5k B
Public keys per KEM variant, straight from the catalog. The McEliece row is the trade: huge keys, tiny ciphertexts, no lattice assumption.
ML-DSA-44 · ML-DSA signature2.4k B
ML-DSA-65 · ML-DSA signature3.3k B
ML-DSA-87 · ML-DSA signature4.6k B
SLH-DSA-SHA2-128s · SLH-DSA signature7.9k B
SLH-DSA-SHA2-128f · SLH-DSA signature17k B
SLH-DSA-SHA2-256s · SLH-DSA signature30k B
SLH-DSA-SHA2-256f · SLH-DSA signature50k B
SLH-DSA-SHAKE-128s / 128f / 192s / 192f / 256s / 256f · SLH-DSA signature7.9k B
Falcon-512 · FN-DSA signature666 B
Falcon-1024 · FN-DSA signature1.3k B
MAYO-1 / MAYO-2 / MAYO-3 / MAYO-5 · MAYO signature321 B
CROSS-RSDP / CROSS-RSDPG (multiple param sets) · CROSS signature13k B
OV-Is / Ip / III / V (multiple param sets) · UOV signature96 B
SNOVA (multiple param sets at NIST levels 1, 3, 5) · SNOVA signature168 B
Signatures per scheme - ML-DSA vs FN-DSA compactness vs SLH-DSA conservatism, drawn to scale.

KEMs

Key Encapsulation Mechanisms (7 families)

Used for key agreement - TLS, KMS-wrapped data keys, vault secret encryption, encrypted object storage.

ML-KEM
FIPS-finalised
Module-Lattice-based Key Encapsulation Mechanism · FIPS 203
NIST's primary post-quantum key encapsulation standard, finalised August 2024 as FIPS 203. QNSI policy sources select ML-KEM across PQC-native tiers and define it for transport and envelope targets; complete production negotiation and envelope execution remain NOT VERIFIED.
lattice-based3 variantsnobleliboqs
/algorithms/ml-kem →
HQC
FIPS-pending
Hamming Quasi-Cyclic Key Encapsulation
NIST-selected code-based backup KEM (selected March 2025); its draft standard is still pending as of October 2026. QNSI offers HQC as an explicit diversity option with liboqs 0.16.0, which updates the implementation to the 20250822 specification; ML-KEM remains the FIPS-final default.
code-based3 variantsliboqs
/algorithms/hqc →
BIKE
non-FIPS
Bit Flipping Key Encapsulation
Code-based KEM finalist (round 4 of NIST PQC standardisation) using QC-MDPC codes. Available in liboqs for QNSI customers seeking additional code-based alternatives.
code-based3 variantsliboqs
/algorithms/bike →
Classic McEliece
non-FIPS
Classic McEliece Code-Based KEM
The original code-based public-key cryptosystem, in continuous study since 1978 - by far the oldest cryptographic assumption in the PQC catalogue. Trades extremely large public keys for the most-studied security assumption available.
code-based1 variantsliboqs
/algorithms/classic-mceliece →
FrodoKEM
non-FIPS
Frodo Key Encapsulation Mechanism
Plain Learning With Errors (LWE) KEM - same lattice family as ML-KEM but without the additional ring or module structure. Larger keys and ciphertexts but built on the most conservative lattice assumption.
lattice-based3 variantsliboqs
/algorithms/frodokem →
NTRU
non-FIPS
Number Theoretic Research Unit Cryptosystem
One of the oldest lattice-based KEMs, in continuous study since 1996. NTRU was a NIST PQC finalist but not selected for FIPS standardisation in favour of ML-KEM.
lattice-based2 variantsliboqs
/algorithms/ntru →
NTRU Prime
non-FIPS
NTRU Prime (Streamlined / Light NTRU Prime)
NTRU variant designed to use a prime-degree ring polynomial, removing certain structural concerns. Notably deployed in OpenSSH's default post-quantum key exchange.
lattice-based1 variantsliboqs
/algorithms/ntru-prime →

Signatures

Digital Signature Algorithms (7 families)

Used for authenticity and non-repudiation - JWT signing, audit-log Merkle root signing, code signing, SPIFFE SVID signing.

ML-DSA
FIPS-finalised
Module-Lattice-based Digital Signature Algorithm · FIPS 204
NIST's primary post-quantum digital signature standard, finalised August 2024 as FIPS 204. ML-DSA powers JWT signing, audit-log integrity, code-signing, and authn token issuance across QNSI.
lattice-based3 variantsnobleliboqs
/algorithms/ml-dsa →
SLH-DSA
FIPS-finalised
Stateless Hash-based Digital Signature Algorithm · FIPS 205
NIST's hash-based digital signature standard, finalised August 2024 as FIPS 205. SLH-DSA's security rests only on the hardness of finding hash function preimages - the most conservative assumption available - making it the natural choice for long-archival signatures and government-tier policy.
hash-based5 variantsnobleliboqs
/algorithms/slh-dsa →
FN-DSA
FIPS-pending
FFT-based NTRU Digital Signature Algorithm · FIPS 206 (pending)
NIST's fourth standardised PQC signature scheme, formally FN-DSA under FIPS 206 - the initial public draft is still the current publication as of October 2026. Falcon's signatures are the most compact of the lattice-based PQC schemes, making it preferred for size-constrained transport.
lattice-based2 variantsliboqs
/algorithms/fn-dsa →
MAYO
non-FIPS
Multivariate Quadratic Signatures (MAYO)
Multivariate quadratic signature scheme advanced to Round 3 of NIST's additional digital signature process (NIST IR 8610, October 2026). Short signatures and small public keys; trades signing speed against parameter size.
multivariate1 variantsliboqs
/algorithms/mayo →
CROSS
non-FIPS
Codes and Restricted Objects Signature Scheme
Code-based signature using the MPC-in-the-head paradigm over restricted syndrome decoding. Reached Round 2 of NIST's additional digital signature process but was not among the nine candidates advanced to Round 3 (October 2026), so it remains research and diversity only.
code-based1 variantsliboqs
/algorithms/cross →
UOV
non-FIPS
Unbalanced Oil and Vinegar Signatures
Multivariate signature scheme based on the Unbalanced Oil and Vinegar (UOV) construction, one of the longest-studied multivariate schemes. Advanced to Round 3 of NIST's additional digital signature process (NIST IR 8610, October 2026).
multivariate1 variantsliboqs
/algorithms/uov →
SNOVA
non-FIPS
Simple Noncommutative-ring-based UOV Algorithm
Multivariate signature scheme using a non-commutative ring structure to reduce public-key size relative to plain UOV. Advanced to Round 3 of NIST's additional digital signature process (NIST IR 8610, October 2026).
multivariate1 variantsliboqs
/algorithms/snova →