Deploy ollama + openweb-ui: run your own open models, own your (free) tokens

Deploy and Host Ollama + Open WebUI

Run your own private AI chat — Ollama serving 100+ open models (Llama, Qwen, Gemma, Mistral, DeepSeek, Phi and more) with Open WebUI on top: a polished, self-hosted ChatGPT-style interface. No API keys, no rate limits, no per-token bills — your prompts never leave your Railway instance.

🎯 Stop renting tokens. Own them. With closed-model APIs you pay for every token, forever. Here, tokens are unlimited and free — you only pay for infrastructure.

🚀 Getting Started with Ollama + Open WebUI on Railway | Deployment Guide

Once your Railway deploy is live, open your Open WebUI URL — the first account you create automatically becomes the admin.

Step 1: Create Your Admin Account

Open your service URL and sign up. The first registered user gets full admin rights over models, users, and settings.

Step 2: Pull Your First Model

Go to Admin Panel → Settings → Models → Pull a model and pull any model from ollama.com by name, for example llama3.1:8b or qwen2.5:7b. Models download straight to your persistent volume, so they survive redeploys. You can also pull curated, community-tuned configs from ollama.com/oamazonasgabriel — optimized for consumer-hardware-class resources.

Step 3: Chat

Pick the model in the top-left dropdown and start chatting. Switch models mid-conversation, run side-by-side comparisons, or attach documents for RAG-powered answers.

Step 4: Connect Your Apps (Optional)

Generate an API key in Settings → Account → API Keys and point any OpenAI-compatible tool at your own endpoint. Your apps get an AI backend with zero per-token billing.

About Hosting Ollama + Open WebUI 📖

This template deploys a production-ready, two-service stack with persistent storage:

• 🦙 Ollama — the model runtime. Pulls and serves open models, exposes an OpenAI-compatible API. Kept private (no public domain) and wired to WebUI over Railway private networking.
• 💬 Open WebUI — the ChatGPT-style interface. Multi-user auth, chat history, markdown + code highlighting, voice input, document uploads (RAG), and web search. This is the only public-facing service.
• 💾 Two persistent volumes — one for pulled models (/root/.ollama), one for users, chats, and uploads (/app/backend/data). Both survive redeploys and updates.

Your models and conversations are 100% yours: nothing is sent to third parties, and there are no token meters running.

Why Deploy Ollama + Open WebUI on Railway ✅

One-click deploy with zero server configuration:

• 🟢 No Docker, driver, or CUDA setup — Railway handles it all
• 💾 Persistent volumes keep models, chats, and accounts across deploys
• 🔐 Private networking between services — Ollama is never exposed publicly
• 📉 Scale down when idle and pay pennies; scale up for heavy sessions
• 🧩 GPU instances available (plan/region dependent) for 10–30× faster inference
• 🌐 Browser-based everything — pull models and manage users without SSH
• 🔄 Redeploy to update images — models and data persist

Common Use Cases 💡

• Private ChatGPT alternative — personal or team AI chat with zero data leaving your server
• Free-token API backend — replace per-token API bills with an OpenAI-compatible endpoint you own
• Document intelligence — upload PDFs and docs, chat with them via built-in RAG
• Team AI workspace — multi-user accounts, shared prompts, per-user history
• Open-model experimentation — benchmark models, tune system prompts, compare answers side-by-side
• Sovereign AI — run open-weight models (Llama, Qwen, Gemma, DeepSeek, Mistral) on infrastructure you control

Dependencies for Ollama + Open WebUI 📦

• Ollama — official ollama/ollama image, model storage persisted at /root/.ollama
• Open WebUI — official ghcr.io/open-webui/open-webui image, data persisted at /app/backend/data
• Railway private networking — WebUI reaches Ollama via OLLAMA_BASE_URL=http://ollama:11434

Dependencies for

Deployment Dependencies

• Ollama: ollama/ollama — Model library — Curated configs
• Open WebUI: open-webui/open-webui — Docs

🖥️ Minimum Hardware Requirements for Ollama + Open WebUI

💡 Start with a 7–8B model — the sweet spot for quality vs. cost. Attach a Railway GPU where available for serious inference speed.

💰 Own Your (Free) Tokens — The Economics

Reality check: you pay for RAM/CPU while running, not tokens. An always-on 8 GB instance is ~$80/mo — heavy API users break even fast. Everyone else can scale down when idle: models reload from the volume in seconds on demand, so light usage costs literal pennies.

🐳 Self-Hosting Ollama + Open WebUI Outside Railway

Prefer your own hardware? The same stack runs anywhere Docker runs:

# 1) Ollama with persistent model storage
docker run -d --name ollama -p 11434:11434 -v ollama:/root/.ollama ollama/ollama

# 2) Open WebUI, pointed at your local Ollama
docker run -d --name open-webui -p 3000:8080 \
  -e OLLAMA_BASE_URL=http://host.docker.internal:11434 \
  --add-host=host.docker.internal:host-gateway \
  ghcr.io/open-webui/open-webui:main

Open http://localhost:3000, create your admin account, and pull models as usual.

❓ FAQ

How do I add models? Admin Panel → Settings → Models → pull by name (qwen2.5:7b, gemma2:9b, deepseek-r1:8b…). Models persist across redeploys on the volume.

Are my chats and prompts private? Yes — everything stays on your Railway instance and volumes. Nothing is sent to third parties.

Can I use my existing OpenAI tools and SDKs? Yes — Open WebUI exposes an OpenAI-compatible API with keys you control. Point any OpenAI client at it.

A model feels slow on CPU. Use a smaller or quantized model (e.g. :4b variants), or deploy in a GPU-enabled region and attach a GPU.

How do I update the stack? Hit redeploy — both images update, and your models, chats, and accounts persist on the volumes.

Can multiple people use it? Yes — every user gets their own account, history, and settings. Admins manage users and default models.