Ethereum Foundation Launches zkAPI to Separate AI Payments From User Identity

MakeCryippreferred on

• The Ethereum Foundation and the Open Anonymity Project launched zkAPI on October 1, 2026, letting users pay for metered APIs with zero-knowledge proofs instead of an identity
• Users deposit credits into an Ethereum vault contract, and software on their device then generates a zero-knowledge proof that authorizes bounded spending without revealing who they are
• zkAPI is live on Ethereum Mainnet, built on a design published on Ethereum Research by Davide Crapis and Vitalik Buterin, with both mainnet and Sepolia testnet deployments available

The Ethereum Foundation published a post on its official blog on October 1, 2026 introducing zkAPI, a privacy-focused payment system built with the Open Anonymity Project that lets users pay for a metered API without being identified to the service they are paying.

The system works through two separate pathways. On the payment side, a user deposits credits, in ether, USDC, or other supported assets, into an Ethereum vault contract, and that balance becomes a private cryptographic note that only the depositor can spend. On the usage side, software running on the user’s own device generates a zero-knowledge proof that authorizes a bounded amount of spending, and a server then issues a short-lived, capped API key without ever learning the identity of the person behind it. Requests then go directly to the underlying AI provider using that temporary key.

The design, authored by Davide Crapis and Vitalik Buterin and first published on Ethereum Research, relies on Groth16 zero-knowledge proofs over the BN254 elliptic curve, Poseidon hashing, and a 32-level Merkle tree to track commitments without exposing the underlying data. Under this architecture, the zkAPI server learns only that a valid payment exists and the total dollar amount spent per session, never who the user is, what they asked the AI model, or which specific deposit funded the request. The AI provider on the other end sees the prompts and responses needed to do its job but never learns who is paying for them, while Ethereum’s public ledger records only the deposits and withdrawals themselves.

The Ethereum Foundation framed the launch as a response to a structural problem in how people currently pay for AI services, where “using AI means handing a running transcript of your thinking to whoever holds the billing relationship.” By separating payment from identity, zkAPI is designed to let users access metered AI APIs without building up a persistent, billing-linked usage profile that a provider could later analyze, sell, or be compelled to disclose.

zkAPI is live now on Ethereum Mainnet, with the Ethereum Foundation publishing GitHub repositories for the underlying code, a browser-based chat interface called OA Chat for trying the system directly, and both mainnet and Sepolia testnet deployments available for developers who want to integrate zkAPI-style payments into their own metered services. The launch adds a concrete, shipped application to the growing body of zero-knowledge research coming out of the Ethereum ecosystem, translating theoretical privacy guarantees into a payment rail developers can plug into today.