The Ethereum Foundation and the OAC project launched the zkAPI. After depositing ETH or USDC, users can obtain temporary keys through zero-knowledge proof to call the AI model to achieve anonymous payments and inject increments into the narrative of AI agent machine payment and ETH settlement layer.
Author: Zack Abrams
Compiled by: Shenchao TechFlow
Deep Trend Introduction: The Ethereum Foundation and the OA project have made the "ZK API Usage Credits" proposed by Vitalik et al. in February into a zkAPI available on the main network: first go to the gold inventory ETH/USDC, and then use the zero-knowledge proof to exchange the temporary key with the quota to call the model, and the payment and identity are separated from the API link. The narrative of the ETH settlement layer and AI agent machine payment are both incremental; however, the gateway may still rely on IP associated requests, and the protocol itself is still marked as experimental.
The system was built with the participation of the Open Anonymity Project and was implemented based on a design published publicly in February by Vitalik Buterin, co-founder of Ethereum, and Davide Crapis, head of dAI at the Ethereum Foundation.
The Ethereum Foundation on Thursday launched zkAPI, a system that allows users to pay for AI models and other pay-as-you-go APIs without revealing their identity, according to a blog post on Thursday.
The foundation stated that zkAPI was jointly built with the open source AI privacy project Open Anonymity (OA) Project and is currently online on the Ethereum mainnet.
“Every AI API call today carries an identity,” the blog reads. "Your API key points to an account, and the account points to a payment method. Every prompt word you send will be merged into the records linked to the two. The service provider can piece together your usage trajectory over the years into a single portrait."
“It seems like a generalization of OA’s non-correlated reasoning while abstracting away payment,” Ken Liu, a Stanford computer science doctoral candidate who participates in the Open Anonymity Project, posted on X on Thursday afternoon. He also wrote, “There should be more infrastructure that uses privacy and sovereignty as its foundation and allows people to truly control themselves.”
How zkAPI works
According to the blog, when using zkAPI, users must first deposit tokens such as ETH or USDC into the treasury contract on Ethereum, and the contract will record the user's balance in the form of a private note. To spend these tokens, software on the user’s device generates a zero-knowledge proof: proving that their request is endorsed by a sufficiently funded note, but without revealing which note or corresponding deposit.
After the zkAPI server verifies the zero-knowledge proof, it will issue a temporary API key with a consumption limit. The user then takes this key and sends the prompt word directly to the AI model service provider; when the key expires, the server will deduct the corresponding usage amount from the user's private balance.
Every time a user pays for AI use, a serial number called "nullifier" will also be announced. “If someone tries to spend the same balance twice, a duplicate nullifier will be generated, exposing the attempt but no other information,” the blog states.
According to the blog, the client can work with existing AI tools. “The client exposes standard OpenAI and Ollama APIs on your local machine,” the blog states. "Existing apps, editors, and chat clients can be used by simply pointing them to localhost."
Use cases and limitations
The blog lists AI chat and agents as top use cases. Other applications named include blockchain RPC queries, image and video generation, VPN bandwidth, and machine-to-machine payments between AI agents.
However, according to the blog, the zkAPI system does not provide network layer anonymity. It is possible for a gateway to correlate requests from fixed IP addresses; if users require greater privacy, the blog states that traffic can be routed through Tor - an anonymous network that relays traffic through multiple servers and is also commonly used by so-called "dark web" sites.
Conversations may also be reconnected through prompt word content, such as personal details, writing style, or conversation history. The zkAPI project's GitHub repository describes the protocol as experimental.
Design co-presented by Vitalik Buterin
The implementation of zkAPI is the "ZK API Usage Credits" design published by Davide Crapis and Ethereum co-founder Vitalik Buterin on the Ethereum Research forum on February 11.
Crapis leads the Ethereum Foundation’s dAI team, which was launched in September 2025 to advance Ethereum’s capabilities as an AI settlement and coordination layer. The team also developed the AI agent identity standard ERC-8004, which went live on the mainnet in January.
The blog announcing zkAPI was written by Vittorio Rivabella from the dAI team.
In an X post in February, Buterin listed cryptographic payment mechanisms for AI services as one of Ethereum’s near- and mid-term visions for the role of AI.