AI Inference Lab

Short-lived OpenAI-compatible inference endpoints for experiments, benchmarks, and agent workloads.

Operational Note

Tunnels are provisioned on demand. This landing page also acts as a smoke test for the Inference Lab domains.

How It Works

  • Client connects over HTTPS
  • Traefik terminates TLS and routes by hostname
  • The inference-lab network reaches an SSH tunnel container
  • The tunnel forwards requests to the model provider

Endpoint Pattern

Model endpoints are exposed as named subdomains.

https://<name>.inference-lab.embedded-focus.comhttps://<name>.inference-lab.gitops.at