Use LLMs withPrivacy
Route across top models with encrypted prompts, secure enclaves, and verifiable execution by default.
Zero-Knowledge Privacy
Your prompts never leave your machine in plaintext. ZK proofs verify inference without exposing inputs.
Learn MoreAny Model, One API
Route to GPT-4, Claude, Mistral, and more through a single OpenAI-compatible endpoint.
Learn MorePay-As-You-Go Credits
Credits never expire. Buy once, use across every model and provider with no hidden fees.
Learn MoreOn-Chain Attestation
Every inference is attested on the 0G chain — cryptographic proof your AI ran correctly.
Learn MorePrivate AI inference.
Built for speed and trust.
Encrypted Prompt Pipeline
Prompts are encrypted on the client and decrypted only inside TEE execution. Providers process jobs without seeing raw user input.
Smart Provider Routing
Route each request to the best available provider based on capability, latency, and price for more consistent user experience.
Verifiable Attestation
Every inference can produce attestations tied to runtime and model integrity, anchored for independent verification.
Escrow-Based Settlement
Usage-based payments settle through escrow with provider stake and proof checks, reducing trust risk for both sides.
Three steps.
Private inference, verified end to end.
1import { ZKai } from "zkai";23const client = new ZKai({4 apiKey: process.env.ZKAI_API_KEY!,5 walletKey: process.env.ZKAI_WALLET_KEY!,6 model: "qwen2.5-1.5b",7});
Plugs into your existing AI stack.
OpenAI-compatible clients, TEE runtimes, model routing, and 0G chain settlement in one pipeline.
Trust is
non-negotiable.
ZKai is built for private AI by default: encrypted prompts, TEE execution, and verifiable settlement across the entire inference lifecycle.
TEE-Isolated Execution
Inference runs inside a Trusted Execution Environment so prompts are decrypted only in enclave memory.
End-to-End Prompt Encryption
Client-side encryption protects inputs and outputs in transit; providers process ciphertext until enclave decryption.
Verifiable Attestation
Every job includes attestations tied to model and runtime integrity, with hashes anchored on the 0G chain.
Escrow + Provider Stake
Usage is settled through escrow with provider staking, so payment and execution integrity are enforced together.
Ready to ship
private AI at scale?
Launch with ZKai using OpenAI-compatible APIs, verifiable attestation, and usage-based settlement on the 0G chain.
Questions