Kaspa Accelerates Its Quantum Shielding with Zero-Knowledge Proofs

• The heaviest signature, which took over 28 minutes to verify, has been reduced to just over 16 seconds.
• The function currently runs on Kaspa's testnet, not on the main network.

A sufficiently powerful quantum computer could, in theory, break the cryptography that currently protects Bitcoin and cryptocurrencies.

This is not an imminent threat nor completely resolved, but it is one that the industry is taking seriously: every bitcoin (BTC), every ether (ETH) transaction, or any network based on elliptic curve cryptography depends on certain mathematical problems remaining, for now, impossible to solve for a conventional computer.

In contrast, an advanced quantum computer could change that equation, and with it, the security of the private keys that control the funds.

That is why the U.S. National Institute of Standards and Technology (NIST) standardized three families of digital signatures designed to withstand that scenario: ML-DSA, SLH-DSA, and a variant of Falcon known as FN-DSA, the latter still in draft phase.

The practical problem is that these post-quantum signatures are much heavier than the current ones: they take up more space and take longer to verify, making them less viable for block networks with limited space and time.

Kii claims to have resolved the bottleneck with ZK proofs.

Kii worked on this problem, the company behind a suite of institutional products built on the Kaspa network.

On July 22, 2026, the Kaspa Kii X account announced that it claims to have verified the three NIST post-quantum signatures within the network in seconds, something that previously took minutes or even half an hour.

The key is not to attach the full signature to the chain, but rather a zero-knowledge proof (ZK) that certifies that the verification of that signature was correct, without repeating the entire calculation at each node.

To generate it, Kii uses the zkVM from RISC Zero, an environment that executes the verification code and produces a succinct STARK proof, compact enough to be anchored on-chain.

According to the figures Kii published in its code repository, run on an RTX 4090 GPU (average of five runs):

• Falcon-512: 1.91 seconds
• ML-DSA-44: 5.35 seconds
• SLH-DSA-128s: 16.3 seconds

The most notable jump is that of SLH-DSA, a scheme based on hash functions that, without this method, took over 28 minutes to verify on a conventional processor. Kii's own repository attributes this improvement to a SHA-256 accelerator included in RISC Zero.

Kii commissioned an independent execution of these tests to GhostProver, the ZK testing infrastructure of the NTH MOMENT project, and also published that repository along with the test artifacts for anyone to audit.

However, it is worth clarifying that upon reviewing that second repository, the total times are not identical to those Kii presents as independent figures: NTH MOMENT reports 2.47 seconds for Falcon-512 and 5.89 seconds for ML-DSA-44, values somewhat different from the 1.91 and 5.35 seconds reported by Kii. Only the SLH-DSA figure (16.22 seconds) matches almost exactly.

Both repositories also use different versions of RISC Zero (3.0.6 in the independent run versus 3.0.5 in Kii's benchmark), a difference that could explain part of the gap, although neither party details it.

Kii's own repository describes the project as pre-production, without certification or guarantee of being ready for production.

The function that anchors these tests on-chain corresponds to a proposed improvement identified as KIP-16, with the tag tag-0x21, which for now only runs on Kaspa's testnet-10, not on the main network.

It is worth clarifying that Kii stands for Kaspa Industrial Initiative in English. It is an organization that seeks to develop commercial products focused on payments, institutional activities, and commercial traceability in the Kaspa ecosystem.

For its part, Kaspa is a cryptocurrency network based on proof of work (PoW) that implements a protocol called GHOSTDAG, which allows for fast confirmation times in mining systems.