The Ethereum Foundation and Open Anonymity Project have launched zkAPI on Ethereum mainnet, allowing users and autonomous software to pay for metered AI services without attaching a billing identity to each request.
Users deposit ETH, USDC or another supported asset into an Ethereum vault. The deposit creates a private note that can authorize bounded API usage through zero-knowledge proofs, according to the Ethereum Foundation’s October 1 technical announcement. Ethereum Foundation Blog
The payment server verifies that a funded note can cover a session and issues a short-lived API key with a dollar spending cap. The AI provider receives the prompts and responses but does not learn the payer’s billing identity. The payment server sees a valid spend but not the prompt.
The design uses Merkle-tree commitments, nullifiers to prevent double spending and Groth16 proofs. Signed usage receipts settle the final bill after a session, avoiding one onchain payment or proof for every model call.
Funds remain in an Ethereum contract rather than a conventional prepaid company account. Users can close their private balance and withdraw onchain even if the zkAPI server stops operating.
The privacy boundary is limited. The AI provider still sees the request content and network metadata such as an IP address. Writing style, personal details, reused history and timing can also link sessions. zkAPI separates billing identity from usage, but it does not make prompts anonymous.
TOKEN RECON ASSESSMENT
zkAPI is a working infrastructure release, not an AI-branded token narrative. It addresses a specific problem for autonomous agents and sensitive API usage: paying continuously without creating a permanent billing profile.
The next evidence threshold is production adoption. Meaningful signals would include independent providers accepting its proofs, measurable mainnet deposits, audited contracts and autonomous applications using bounded spending without human intervention.
