Your own sovereign AI provider with Infersec Self-Hosted

  • Home
  • / Your own sovereign AI provider with Infersec Self-Hosted
Your own sovereign AI provider with Infersec Self-Hosted

If you haven't been watching the open weight, freely-available model space - now's the time to catch up. We've seen a ferocious push in releasing new models, especially from China, but also including a number of smaller models from the US and other countries - they're all super interesting, and most have their place in a particular workflow you may be running or planning to build.

Infersec's cloud hosting option for connecting your local compute to a production-ready AI endpoint has been available for some time now, but we recognise that for large-scale, data-sensitive operations this may not be enough. There are of course many use cases for AI platforms that require a strict, accountable routine for handling highly-sensitive data, like in the health and law sectors, where a cloud-hosted option simply won't cut it.

A self-hosted Infersec

This is why we've built our self-hosted option: a Docker-based installation of Infersec that runs on-premises or in your own private cloud, where the data stays entirely in your hands and never reaches our cloud offering. You get the full feature set Infersec ships - routing, authentication, monitoring, MCP tool calling - with total control of how traffic is routed and how your endpoints are consumed. The self-hosted docs cover the full setup.

Point Conduit at a self-hosted instance

Conduit, the application that connects your compute to Infersec, points just as easily at a private self-hosted instance as at the cloud, so any workflow you've built on the hosted version migrates across without rework. Scaling out is fine too - each self-hosted license covers a single deployment, so stand up a second instance with its own key rather than sharing one across hosts.

Observability on your own stack

Self-hosted suits private clouds with GPU-equipped instances nicely - run our stack end-to-end and become your own sovereign AI provider. Just like the cloud version, the self-hosted variant emits OpenTelemetry metrics, so you can monitor it with the tooling you already run: Grafana, InfluxDB, Kibana, Elasticsearch, OpenSearch. Pricing is annual per tier - see the pricing page for the current caps.

Sign up today for early access to the platform and to provision a self-hosted license. Licenses ship with a trial period so you can explore the product before committing - and since the cloud version is almost feature-identical, a trial there is often the fastest way to poke around.

Cover image generated with AI.