Nikolas Sargeant
TL;DR
-
The Ethereum Foundation and Open Anonymity Project launched zkAPI on Ethereum mainnet.
-
Users can fund private balances with tokens such as ETH or USDC and obtain temporary, spending-capped API keys.
-
Zero-knowledge proofs verify available funds without identifying the deposit backing a request.
The Ethereum Foundation has introduced zkAPI, a system designed to separate payments for AI models and other metered services from users’ account identities.
Built with the Open Anonymity Project, the system is live on Ethereum mainnet, according to the Foundation’s Thursday announcement.
Its goal is to reduce the connection between a payment, an API account, and a persistent record of usage. However, zkAPI does not provide complete anonymity: network information and the contents of prompts can still reveal or connect a user’s activity.
Private Credits Separate Funding From Requests
Using zkAPI begins with depositing tokens, such as ETH or USDC, into an Ethereum vault contract.
The system records the balance as a private note. When a user wants to access a service, software running on their device generates a zero-knowledge proof showing that the request is backed by sufficient funds.
The proof does not disclose which note or original deposit provides those funds.
A zkAPI server verifies the proof and issues a temporary API key with a spending limit. The user then sends prompts directly to the AI provider using that key.
When the key expires, the server deducts the cost of usage from the private balance. The design aims to let providers verify payment without attaching every request to the same conventional billing account.
zkAPI uses a serial number called a nullifier to prevent users from spending the same balance more than once.
A nullifier is published when payment occurs. Attempting to reuse the same balance produces a duplicate, allowing the system to detect the double-spending attempt without revealing the underlying private note.
This addresses a central requirement for private payments: a service must know that credit is valid and has not already been spent, even when it cannot identify the deposit behind it.
Temporary API keys may still group activity within a session. The system’s privacy objective concerns reducing links to funding identities and across usage, rather than making each prompt invisible to its provider.
Existing AI Tools Can Connect Through Local APIs
The client exposes interfaces compatible with the standard OpenAI and Ollama APIs on the user’s machine.
Existing applications, editors and chat clients can connect by pointing their requests to localhost, according to the Foundation.
AI chat and agents are the initial use cases. The announcement also identifies blockchain RPC queries, image and video generation, VPN bandwidth and machine-to-machine payments as potential applications.
These services share a common requirement: charging for measured usage. zkAPI provides a payment mechanism without requiring that every purchase be associated with a persistent customer identity.
The system does not provide network anonymity. A gateway could correlate requests arriving from a stable IP address. Users seeking stronger network privacy may route traffic through Tor, the Foundation said.
Prompt content presents another limitation. Personal details, distinctive writing patterns and reused conversation history can connect sessions even when the payment mechanism is unlinkable.
The AI provider still receives the prompts needed to deliver its service. Private payment therefore does not mean private processing of their contents.
The project’s GitHub repository describes zkAPI as experimental, despite its mainnet availability.
Buterin-Coauthored Design Underpins the Launch
zkAPI implements the ZK API Usage Credits design published in February by Davide Crapis and Ethereum co-founder Vitalik Buterin.
Crapis leads the Ethereum Foundation’s dAI team, formed in September 2025 to develop Ethereum’s role as a settlement and coordination layer for AI.
The team also developed ERC-8004, an AI-agent identity standard launched on mainnet in January.
With zkAPI, the Foundation is adding a complementary payment tool: one intended to verify usage credits while limiting the identity information attached to them.
