Open WebUI CLI Cheatsheet - open-webui Self-hosted AI Platform Full Reference

Open WebUI solves the problem of teams needing models without everyone managing separate vendor consoles: it aggregates Ollama and any OpenAI-compatible service behind one interface with user accounts, an admin panel, knowledge bases and conversation history, so model access is issued centrally instead of via personal API keys. This sheet covers installation and launch, updating and maintenance, and deployment and configuration. Two operational details decide whether the deployment survives. First, the data directory must be set explicitly - launched via a temporary runner, chats, user settings and knowledge-base indexes disappear when the process exits, and that loss usually surfaces only when someone reports their history is gone. Second, concurrency during upgrades: running database migrations with multiple replicas or workers in parallel can corrupt the store, so the correct sequence is scale down to a single replica, complete the migration, then scale back out. These two are the most common real failure sources in self-hosted deployments; most other settings can be tuned later.