Ethereum's zkAPI Goes Live on Mainnet, Bringing Private Payments to AI Services

The Ethereum Foundation turned a February research proposal into production code on October 2, 2026. zkAPI is now live on Ethereum mainnet, offering privacy-preserving prepaid access to AI models and other metered APIs through zero-knowledge proofs.

Built with the Open Anonymity Project, zkAPI lets users pay for services without revealing which deposit funded their requests. For the Web3 and AI intersection, it addresses a growing tension: powerful AI tools require payment infrastructure, but users increasingly want financial privacy.

How zkAPI works

Users deposit tokens — ETH or USDC — into a vault contract on Ethereum. The deposit creates a private note recording the balance. When spending, client software generates a zero-knowledge proof demonstrating sufficient funds without revealing which note or deposit backs the transaction.

The system issues short-lived API keys with predefined spending limits. Prompts go directly to the AI provider. Usage settles separately through the payment layer. The API provider and payment server cannot link a request to a specific payer.

The Ethereum Foundation also released a local client, SDK, and browser-based AI chat implementation to demonstrate the flow.

Origins in research

zkAPI implements "ZK API Usage Credits," a design Ethereum Foundation researcher Davide Crapis and co-founder Vitalik Buterin published on the Ethereum Research forum in February 2026. The mainnet launch validates the concept with working infrastructure.

Vittorio Rivabella, AI coordinator at the Ethereum Foundation's dAI team, framed immediate use cases around paying for AI without linking payment to identity. Additional applications include blockchain RPC queries, image and video generation, VPN bandwidth, and machine-to-machine payments between AI agents.

What privacy actually means here

zkAPI protects billing identity, not everything. The system does not hide prompt contents or network metadata from providers. Users can still be linked across sessions through IP addresses, timing patterns, or information embedded in requests.

This is an important distinction for anyone evaluating the system. Financial unlinkability is valuable — especially for agent-to-agent micropayments — but it is not full anonymity.

Why it matters for Web3

The crypto industry has spent years building payment rails. zkAPI applies those rails to the fastest-growing consumption category: AI inference. As agents proliferate and API calls multiply, metered billing becomes infrastructure, not a feature.

Machine-to-machine payments between AI agents — listed explicitly as a use case — could become significant if agent economies develop as predicted. Agents need to pay for tools, data, and compute without human intervention at every step.

Adoption questions

Mainnet launch is step one. Developer adoption depends on SDK quality, provider integration, and whether AI companies see privacy-preserving billing as a competitive advantage or compliance complexity.

The Ethereum Foundation has a track record of shipping research into production. zkAPI's success will be measured by whether developers build on it — and whether users choose cryptographic privacy over convenience of credit cards.

For Web3 builders, zkAPI is the clearest signal yet that Ethereum's next chapter includes AI-native payment infrastructure.