Interfață AI proprie cu Open WebUI pe serverul tău

Open WebUI is an open-source web interface for AI models (LLMs). It looks like ChatGPT, but runs on your own server. Supports local models via Ollama (Llama, Mistral, Gemma) and cloud models via the OpenAI or Anthropic API.

Any Ollama-compatible model: Llama 3.1/3.3, Mistral, Gemma, Phi-3, Qwen, DeepSeek, and dozens more. Download models directly from the Open WebUI interface. Bigger models need more RAM — the Pro plan (16 GB) allows 13B models, the Business plan (32 GB) supports 30B+ models.

Yes. Open WebUI natively supports the OpenAI and Anthropic APIs. Add the API key in settings and get access to GPT-4o, Claude, or any other cloud model — from the same interface, alongside your local models.

When using local Ollama models, absolutely nobody else sees the conversations — not even HostX. Traffic stays on your server. If you connect the OpenAI API, conversations go through their servers (per the OpenAI ToS).

Yes. Open WebUI has a role-based account system (Admin, User). Create accounts for each colleague, with their own conversation history and configurations. All share the VPS resources.

It depends on the model's size. 7B models (Mistral 7B, Llama 3.2 8B) run comfortably on the Standard plan (8 GB RAM). 13B models need at least 16 GB (Pro plan). The Business plan with 32 GB RAM supports 30B+ models.