what's ezkl?
ezkl is an open-source engine for performing zero-knowledge proofs (zk-SNARKs) on deep learning models and other computational graphs, enabling verifiable machine learning (ZKML). It automates the generation of provers and verifiers from high-level model descriptions like PyTorch, allowing inference to be proven tamper-proof and verified onchain, such as on Ethereum and compatible networks. The system supports CLI and Python bindings for ease of use, with features like remote proving services for speed. It powers applications like cryptoidol.tech for onchain AI judging and integrates with ecosystems like Optimism for fair distribution verification, focusing on trustless AI outputs resistant to tampering or hardware failures.
ezkl proofs now verifiable on zkverify testnet for onchain ML operations.
Experienced a phishing attempt targeting the team where attackers impersonated Kenetic Capital representatives and attempted to deliver malware through binary downloads.
ezkl proofs can now be verified on zkverify testnet for onchain ML operations.
Major performance gap revealed as competitor demonstrates 54x-158x faster proof generation and 500x+ verification speed advantage.
Technical optimization achieves 120% accuracy improvement with minimal proof size increase and faster proving time.
New partnership with Inference Labs launches Subnet 2 Incubator grant for iOS age verification model development.
Partnership with Balancer implements ZK-verified dynamic fees for AMMs with working proof of concept.
Major DeFi protocol Balancer integrates technology for real-time dynamic fee adjustment based on market volatility.
Mentioned as being run by Ritual.