LayerZero has introduced Akita, a lattice-based polynomial commitment scheme designed for post-quantum security in zero-knowledge proving systems. Announced September 9, 2026, Akita is described as the first production-ready primitive of its kind and will first appear in Jolt, the zero-knowledge virtual machine developed with a16z crypto that powers the Zero blockchain.
How Akita Works
Akita is a polynomial commitment scheme, a core component in zero-knowledge proving systems that commits to and proves the correctness of a computation trace. By replacing this component with a quantum-secure version, the entire proving stack gains quantum resistance while the rest of the stack already carries information-theoretic security, according to LayerZero.
The scheme uses lattice-based cryptography, the same mathematical class that the U.S. National Institute of Standards and Technology selected for post-quantum encryption and signatures now being deployed to protect encrypted internet traffic.
Performance Improvements
Akita's design delivers significant efficiency gains compared to existing post-quantum proving methods. Hash-based schemes typically produce proofs of 200kb or larger, while Akita produces proofs between 65kb and 80kb, representing a reduction of two to eight times depending on the application.
Proving speed improves by two to three times compared to Jolt's current performance, and memory usage drops by roughly half. Akita requires only sub-linear memory beyond storing the polynomial itself.
Integration and Future Development
Akita will be built into Jolt from day one through LayerZero's partnership with a16z crypto. LayerZero describes the result as the fastest production-grade zkVM currently offering quantum resistance.
The company positions Akita as an early step toward a fully quantum-secure Zero stack, with additional post-quantum upgrades planned as development moves toward mainnet and beyond. The scheme is open source, and LayerZero noted that any proof system can adopt it. The technical paper and implementation are available on GitHub.
