Private inference — attested Apple Silicon
Uncensored models and community fine-tunes, served from verified Macs — where the machine's owner can't read your prompts, and you can prove it.
Keep your OpenAI client. Reach the long tail — uncensored models, rare fine-tunes — that no hosted API will touch. Your prompts stay private.
Host the models you choose — pull them with your own Hugging Face key — and earn per token. The work runs where you can't be watched, and neither can the prompts.
A hardware key, sealed in a chip that can't be cloned.
SIP on, signed system volume, locked down at the kernel.
Apple's CA signs that this is genuine, unmodified hardware.
If the machine changes state, it drops out of the network.
Only our signed code can serve. It can't read or log a prompt.
Every response traces back to real Apple hardware running code that physically can't see your prompt. You don't take our word for it — click "verify attestation" and the X.509 chain checks in your own browser.
Hosted APIs serve a curated, filtered shortlist. The uncensored models and community fine-tunes people actually want have no API at all. Here, a provider chooses to host them — and you just call them.