[00] Local AI hosting platform
Your hardware
your AI
Infersec turns your own AI-capable machines into a production-grade, OpenAI-compatible API. Routed, metered and monitored — on our cloud, or entirely self-hosted.
$ curl -fsSL https://infersec.ai/conduit.sh | bash
==> Resolving latest Conduit version...
==> Downloading Conduit 1.119.0 (linux-x64)...
==> Detecting hardware and fitting models...
✓ hardware detected — 2× GPU, 48 GB RAM
✓ 3 models fit — endpoint ready: /v1/chat/completions
Private
Your hardware, your models
EU based
European sovereign AI
Cloud or Local
Self hosted or managed hosting
Compatible
OpenAI + Anthropic endpoints
01
Conduit
Self-hostable agent that connects vLLM, llama.cpp and other engine servers to the platform.
02
Compute Source
Your Mac or Linux machine, registered as an online source with live engine, model and health telemetry.
03
Open Weight Models
Serve any open weight model locally, with a curated recommended-models catalog.
04
Inferencing Endpoints
OpenAI-compatible inferencing endpoints, public or private, with composable routing rules.
Curated to run on your hardware
Every model is picked, sized and tested against real consumer hardware - from a 230M that runs on a phone to a 284B MoE flagship.
Metered utilities
Credit-based, pay-as-you-go. Top up via Stripe; charges deduct in real time as sources stay online and tokens flow.