Administration

These pages are for the person who runs a LocalAI instance: users and access, settings, monitoring, proxies, distributed mode and troubleshooting.

Access and settings

• Authentication and authorization: API keys, users, roles and per-user features.
• Runtime settings: change settings in the web interface without a restart.
• Reverse proxy and TLS: put LocalAI behind a TLS proxy.

Watch what the instance does

• Operate overview: where to look first.
• Traffic: usage, failed requests, models, GPU and host.
• Activity: watch, cancel and retry installs.
• Backend monitor: list, pre-load and stop running backends.
• Tracing and Profiling and request diagnostics.

Proxies and middleware

• Middleware: PII filtering and routing
• Cloud passthrough proxy
• MITM proxy for Claude Code and Codex CLI

More than one machine

• Distributed mode
• P2P and federated inference and its API
• MLX distributed

When something fails

• Troubleshooting
• Routing logs in embedded applications