The Ethereum Foundation and the Open Anonymity Project launched zkAPI on October 1, a zero-knowledge protocol that lets users pay for metered APIs, starting with AI inference, without the payment ever being linkable to their identity or their prompts.
"Today the EF and OA announce zkAPI—a new means for private AI," the Ethereum Foundation posted on X. "Picture this: deposit ether into a vault, sign a zk proof, get a fixed amount of private inference. Advances like these help us move past middlemen's requirements that we link our data and identities."
The protocol runs live on Ethereum mainnet today, built on a design originally proposed by Vitalik Buterin and Davide Crapis and published on Ethereum Research. The Open Anonymity team, led by Ken Liu, built out the full implementation.
"I'm excited because more infrastructure should exist where privacy and sovereignty is foundational and people are in control," Liu wrote on X.
Quick updates on the Open Anonymity project (@OpenAnonProject)!
— Ken Liu (@kenziyuliu) October 1, 2026
1. OA-chat is live! You can now signup: https://t.co/gxk3cSzIRL
2. We collaborated with @ethereumfndn (@VitalikButerin, @DavideCrapis, @VittoStack, @curryrasul) to develop zkAPI, a new zero-knowledge protocol for…
What problem zkAPI is actually solving
Every API call tied to a traditional key creates a permanent link between an account, a payment method, and the content of every request sent through it. The Open Anonymity team described the stakes plainly in its zkAPI announcement:
"Prompts are personal. People ask AI models about their health, their finances, their doubts. Under the current model, using AI means handing a running transcript of your thinking to whoever holds the billing relationship."
zkAPI breaks that link by separating payment from content entirely. A user deposits ETH or USDC into a vault contract on Ethereum in one public transaction. From that point forward, the balance exists as what the protocol calls a private note, digital cash that only the depositor's device can spend and that cannot be traced back to the original deposit.
How the cryptography actually prevents double-spending
To authorize a payment, the user's device generates a zero-knowledge proof stating that a funded note covers the spend and has not been used before, without revealing which note, deposit, or person is behind it. Deposits exist as commitments inside a 32-level Merkle tree, so a proof can confirm a note is valid without pointing to a specific entry. Every spend also publishes a nullifier, a one-way serial number derived from the note's secret, which flags any attempt to spend the same balance twice without exposing anything else about the transaction.
Nullifiers are the same mechanism Tornado Cash popularized for Ethereum privacy pools starting in 2019, where a one-way cryptographic output prevents double-spending without revealing which specific deposit is being withdrawn. zkAPI applies that proven double-spend prevention technique to a metered payment context rather than a simple mixing pool, extending a well-tested privacy primitive into a new use case: authorizing bounded API usage rather than a single fixed withdrawal. The underlying cryptography, Groth16 proofs on the BN254 curve with Poseidon hashing, is standard tooling across Ethereum's zero-knowledge ecosystem, used widely enough that the security assumptions behind zkAPI rest on infrastructure already battle-tested in production.
The three-party privacy split that makes this work
The architecture is built so that no single party ever holds both the identity and the content. The zkAPI server confirms a valid payment exists and tracks total dollars spent per session, but never learns who the user is or what they asked. The AI provider sees the prompts and responses needed to run the model, but never learns who paid. Ethereum's public ledger only records deposits, closes, and withdrawals, never what any specific balance was spent on.
This works through short-lived API keys. The zkAPI server verifies a payment proof and mints a fresh key on the spot, capped in dollars and held only in the user's device memory. Prompts travel directly from the device to the AI provider using that key. When the key expires, a signed usage receipt charges the user's private balance for actual consumption, and neither side can alter the bill afterward.
What this means for OA Chat users today
The Open Anonymity Project's own chat client, OA Chat, now offers users two sign-in paths. One option uses a traditional account with a username and passkey, paying through Stripe. The second connects an Ethereum wallet directly to the zkAPI vault, requiring no account at all.
"We shoved all the cryptographic machinery into the browser via WebAssembly so you can use zkAPI directly on the OA chat client," the team wrote in its September 25 post announcing the collaboration.
Compiling zero-knowledge proof generation to WebAssembly so it runs inside a standard browser tab removes a significant technical barrier that has limited zk-based privacy tools to developers comfortable running local software or command-line clients. Making that same cryptography execute client-side in-browser is what lets a non-technical user connect a wallet and get unlinkable inference without installing anything, a meaningful usability improvement over prior zero-knowledge privacy tools that typically required local node setup or command-line proof generation.
Where the privacy guarantees stop
The protocol's own documentation is direct about its limits. zkAPI does not provide network anonymity, meaning a provider can still see a user's IP address and attempt to correlate request timing patterns. The team recommends routing through Tor with a fresh circuit per session for users who need stronger network-level protection. Separately, prompt content itself can act as a fingerprint: if a user reuses the same writing style, personal details, or project documents across sessions, an inference provider could re-link those sessions even though the payment layer stays fully anonymous.
Beyond AI chat, the team said the same client and contracts can front any metered service, including blockchain RPC queries, image and video generation, VPN bandwidth, and machine-to-machine payments between autonomous agents. Developers can explore the full implementation through the OpenAnonymity/zkapi GitHub repository, and the live ZkApiVault contract is already deployed on Ethereum mainnet alongside a Sepolia testnet version for experimentation.
Gunel Ismayilova
Disclaimer: All materials on this site are for informational purposes only. None of the material should be interpreted as investment advice. Cryptocurrency investments involve significant risk of loss. Please note that, despite the nature of much of the material created and hosted on this website, HODL FM operates as a media and informational platform, not a provider of financial advisory services. The opinions of authors and other contributors are their own and should not be taken as financial advice. If you require advice, HODL FM strongly recommends contacting a qualified industry professional.
