Quick answer: Ollama on VPS = local AI models without cloud. Install: curl -fsSL https://ollama.com/install.sh | sh, then ollama pull llama3.
Related
Self-host on VPS: Order VPS · NexoraHost · nexorahost.com (Frankfurt · GDPR · NVMe)
Frequently asked questions (FAQ)
Which VPS for Ollama?
7B models: 8 GB RAM minimum. 13B: 16 GB. 70B quantized: 32–64 GB. GPU VPS optional for speed, CPU works too.
Ollama vs OpenAI API?
Ollama: one-time VPS cost, data stays local, GDPR-friendly. OpenAI: pay-per-token, best quality, data in US cloud.
Which models with Ollama?
llama3, mistral, codellama, phi3, gemma – ollama pull llama3. Quantized (Q4) saves RAM.
Use Ollama API from another app?
Yes – http://localhost:11434/api/generate. Expose externally only via reverse proxy + auth.
Ollama with Open WebUI?
Open WebUI = ChatGPT-like interface for Ollama. Docker: both containers on same network.