Newsroom
Breaking
NewsLyfe Breaking News ticker test — preview mode·Ceci est un test — aperçu du bandeau Breaking News·Dies ist ein Test — Vorschau des Breaking-News-Tickers·Esto es una prueba — vista previa del ticker de noticias·这是测试 — 突发新闻滚动条预览·이것은 테스트입니다 — 속보 티커 미리보기·Это тест — предварительный просмотр ленты срочных новостей·Це тест — попередній перегляд стрічки термінових новин·Dit is een test — voorbeeld van de Breaking News-balk·هذا اختبار — معاينة شريط الأخبار العاجلة
Guide/AI/Ollama (Local)

Ollama (Local)

100% private · Local AI setup guide

What is Ollama?

Ollama lets you run powerful open-source AI models directly on your own computer — completely offline, with zero data sent to any server. It's the most private option available, and it's free to use with no API costs. The trade-off is that it requires a capable machine and some initial setup.

🔒
Maximum privacy. When using Ollama, your news articles and summaries never leave your device. Ideal if you read sensitive or confidential content.

System Requirements

🖥️
Operating System

macOS 11+, Windows 10+, or Linux. Apple Silicon Macs (M1/M2/M3) work especially well.

💾
RAM

Minimum 8GB. 16GB recommended for larger models. The model must fit in memory.

💿
Disk Space

3–8GB per model. Models are downloaded once and stored locally.

Setup Guide

01

Download and install Ollama

Go to ollama.ai/download and download for your OS. Run the installer.

02

Download a model

Open a terminal and run:

ollama pull llama3.2

This downloads the Llama 3.2 model (~2GB). Other good options: mistral, gemma2, qwen2.5.

03

Start the Ollama server

Ollama runs as a background service. On macOS/Windows it starts automatically. On Linux, run:

ollama serve
04

Configure NewsLyfe

Open NewsLyfe → Settings → AI → Select Ollama. Enter your local server URL (default: http://localhost:11434) and the model name (e.g. llama3.2).

💡
Recommended models for news summarization:
llama3.2 (fast, 2GB) — mistral (balanced, 4GB) — gemma2:9b (high quality, 5GB)

Cost

API cost$0forever

No API costs, no subscriptions, no usage limits. You only pay for electricity.

Hardware costYour PCone-time

Runs on hardware you already own. No cloud required.

Troubleshooting

Connection Refused

Ollama isn't running. On Linux, run ollama serve in a terminal. On macOS/Windows, check that the Ollama app is open in your system tray.

404

Model Not Found

The model name in NewsLyfe doesn't match what's installed. Run ollama list in your terminal to see installed models, and update the name in Settings.

🐢

Slow Responses

The model is too large for your hardware. Try a smaller model: ollama pull llama3.2 (2GB) runs well on most machines. Apple Silicon Macs are significantly faster than Intel.

Privacy

🔒
Zero data leaves your device. The AI runs entirely locally. No article content, no summaries, and no usage data is sent anywhere.
📡
Works offline. Once the model is downloaded, Ollama works without an internet connection.
🔑
No API key needed. There's no authentication — just point NewsLyfe at your local Ollama server URL.