Local AIPrivacyOffline modelsDeveloper APIs
Ollama
Open-source software to download and run open LLMs locally — with a REST API compatible with many AI apps.
Ratings
Editorial scores for beginner task fit (1–10). Not lab benchmarks — verify pricing and features on official sites.
Pros
- Runs models on your machine
- Large open model library
- OpenAI-compatible local API on port 11434
Cons
- Needs capable RAM or GPU for good speed
- Not a polished chat website for total beginners
- Model quality varies by hardware
Best use cases
- Run Llama or Qwen locally
- Private offline chat
- Connect Cursor or Obsidian plugins to a local model
- Test models without cloud API bills
- Launch agent tools with local inference
Best when
- You want AI answers without sending data to cloud providers.
- You can install desktop software and accept hardware limits.
- You need a local API for dev tools, agents, or self-hosted workflows.
Not ideal when
- You want the simplest browser chat with zero setup (use ChatGPT or Gemini).
- Your device has very little RAM or no GPU and feels too slow.
- You need live web research, slide design, or image generation in one app.
How it compares
- Infrastructure layer — unlike ChatGPT, which is a hosted product.
- Often paired with Cursor, Obsidian plugins, or OpenClaw instead of replacing them.
- Different from DeepSeek cloud — Ollama runs models you choose on your machine.
Caution
Local does not mean risk-free — still review what apps can access your Ollama API and which models you expose.
