Benchmarks Docs Pricing Blog About
Log In Get API Key

H33 vs Zama: Production FHE With ZK + Post-Quantum Signatures

Zama builds FHE libraries. H33 runs FHE in production at 2.17 million operations per second. Four FHE engines, ZK-STARK proofs, and Dilithium signatures — delivered as a single REST API call. No parameter tuning. No cryptography PhD required.

Get Free API Key See Benchmarks
Feature Comparison

Zama vs H33 — Side by Side

How an open-source FHE library compares to a production FHE platform with integrated ZK and post-quantum cryptography.

Feature Zama H33
FHE Scheme TFHE (single scheme) BFV, CKKS, BFV-32, FHE-IQ — 4 engines
Production Readiness Library — you build the infrastructure REST API — production-ready with monitoring
Throughput Varies by implementation 2.17M ops/sec sustained (Graviton4, 96 workers)
Biometric Matching Manual implementation required 939µs per batch (32 users/ciphertext)
ZK Proof Integration Not included ZK-STARKs with SHA3-256 — built in
Post-Quantum Signatures Not included ML-DSA (Dilithium) + ML-KEM (Kyber) — built in
Pricing Open source (BSD) — self-hosted costs $0.033 per operation — volume discounts available
Deployment Model Self-hosted library (Rust / Python) Managed REST API — one call, full stack
Why Teams Switch

Four Reasons to Choose H33 Over Zama

4 Engines

Four FHE Schemes, One API

Zama offers TFHE — a single FHE scheme optimized for boolean circuits. H33 provides BFV for batched integer arithmetic, CKKS for approximate floating-point computation, BFV-32 for lightweight operations, and FHE-IQ for adaptive workloads. Choose the right engine for your use case without switching libraries.

REST API

Production Platform, Not a Library

Zama gives you building blocks. H33 gives you a production service. No parameter tuning, no noise budget management, no key rotation infrastructure to build. Send a request, get cryptographically verified results. Deployed and monitored on AWS Graviton4 with sub-millisecond FHE batch latency.

FHE+ZK+PQ

Combined Cryptographic Stack

Zama provides FHE computation only. H33 combines FHE biometric matching, ZK-STARK zero-knowledge proofs, and Dilithium/Kyber post-quantum signatures into a single API call. Every authentication is encrypted, proven, and quantum-safe — with no additional integrations required.

2.17M/sec

Internet-Scale Throughput

H33 processes 2.17 million operations per second sustained on production hardware. Each operation includes FHE computation (939µs per 32-user batch), a ZK-STARK proof, and a Dilithium digital signature — approximately 38.5 microseconds per individual authentication. Benchmarked, reproducible, and battle-tested.

Developer Experience

Code Comparison

FHE with a library vs. FHE with an API — the developer experience difference.

Zama Concrete — FHE Library
# Zama: manage parameters, keys, circuits
from concrete import fhe

@fhe.compiler({"x": "encrypted"})
def match_biometric(x):
    return (x - template) ** 2

circuit = match_biometric.compile(inputset)
circuit.keys.generate()

encrypted = circuit.encrypt(biometric_data)
result = circuit.run(encrypted)
decrypted = circuit.decrypt(result)
# You manage: parameters, noise, keys,
# deployment, scaling, monitoring...
H33 — Production FHE API
// H33: one call, full cryptographic stack
const result = await h33.authenticate({
  biometric: capturedTemplate,
  securityLevel: 'h33-128',
  mode: 'standard'
});

// result.verified     → true / false
// result.attestation  → Dilithium-signed proof
// result.zkProof      → ZK-STARK verification
// result.fheEngine    → 'BFV' (auto-selected)
//
// FHE batch: ~939µs (32 users)
// ZK proof + PQ attestation included
// No parameters, no keys, no circuits
FAQ

Frequently Asked Questions

What is the difference between H33 and Zama?
Zama provides open-source FHE libraries (TFHE-rs in Rust, Concrete in Python) that developers integrate into their own applications. H33 is a production FHE platform delivered as a REST API. H33 includes four FHE engines (BFV, CKKS, BFV-32, FHE-IQ), integrated ZK-STARK proofs, and post-quantum Dilithium/Kyber signatures — all accessible through a single API call. Zama focuses exclusively on FHE computation and does not include ZK proofs or post-quantum signatures.
Can H33 replace Zama in my application?
If you are using Zama for encrypted computation tasks like biometric matching, credential verification, or encrypted data processing, H33 can replace it with a simpler integration model. Instead of managing FHE parameters, key generation, and encryption/decryption pipelines yourself, you make a REST API call. H33 handles the FHE computation, ZK proof generation, and post-quantum attestation server-side. For custom FHE research or novel circuit design, Zama's library approach gives more low-level control.
Which FHE scheme is faster — TFHE or BFV?
It depends on the workload. TFHE (used by Zama) excels at boolean circuits and bit-level operations with fast bootstrapping. BFV (used by H33) excels at batched integer arithmetic — H33 packs 32 biometric templates into a single ciphertext and processes them in 939 microseconds. For authentication and matching workloads, BFV with SIMD batching is significantly faster. H33 also offers CKKS for approximate arithmetic, BFV-32 for lightweight operations, and FHE-IQ for adaptive workloads.
Does H33 require FHE or cryptography expertise?
No. H33 abstracts all cryptographic complexity behind a REST API. You do not need to choose FHE parameters, manage noise budgets, configure polynomial rings, or understand lattice cryptography. You send data to the API and receive verified results with post-quantum attestation. Zama requires understanding of FHE concepts including parameter selection, noise management, circuit depth, and bootstrapping strategies.
How does H33 pricing compare to Zama?
Zama's libraries (TFHE-rs and Concrete) are open-source under BSD license — free to use, but you pay for your own infrastructure, engineering time, and operational overhead of running FHE workloads. H33 charges $0.033 per operation with volume discounts. The trade-off is self-hosted complexity versus managed API simplicity. For teams without dedicated cryptography engineers, H33's per-operation pricing is often more cost-effective than the engineering investment required to productionize Zama.
Can I use custom FHE parameters with H33?
H33 offers four pre-configured FHE tiers: H33-128 (BFV, N=4096, 128-bit security), H33-256 (BFV, N=8192, 256-bit security), H33-CKKS (approximate arithmetic for ML workloads), and H33-BFV32 (lightweight 32-bit operations). These cover the vast majority of production use cases. For custom parameter sets or novel FHE circuits, Zama's library approach provides more flexibility. H33 prioritizes production readiness and throughput over parameter customization.

Production FHE Without the Library Overhead

One API call. Four FHE engines. ZK proofs and post-quantum signatures included.
Get Free API Key Read Documentation
1,000 free operations per month. No credit card required.