OffRouter

OFFROUTER

OffRouter is a private AI inference service for software agents and developers. It provides an OpenAI-compatible API, so an existing application can switch to it by changing the base URL. The models it serves, 21 open models including DeepSeek, Qwen, Kimi, GLM, Gemma and Llama, run inside confidential GPU enclaves (Intel TDX with NVIDIA confidential computing), and prompts and answers are not stored or logged. Each response includes a receipt signed inside the enclave that can be checked against the enclave's hardware attestation. OffRouter does not require an account. Users pay per request through the x402 protocol or with a prepaid key, in USD stablecoins on Base, Robinhood Chain and Arc. OFFROUTER is the utility token of the service, an ERC-20 token on Robinhood Chain with a fixed supply of 1,000,000,000. Wallets holding OFFROUTER receive a discount on inference (5%, 10% or 20% at the Silver, Gold and Platinum tiers) and a weekly allowance of free LLM tokens that resets every Monday. The tier is read from the wallet balance at the time of each request, with no staking or locking.