One command to run open models on your machine
Ollama is an MIT local model runtime. Pull weights, chat in the terminal, and use an OpenAI-compatible API (default http://localhost:11434). The runtime is free; each model's license is separate.
Updated 2026-09-04. Same-day snapshot of the official repo.
How to Get Started with Ollama
3 steps to your first local model.
Install
Official README: install script on macOS/Linux; installer or PowerShell on Windows.
curl -fsSL https://ollama.com/install.sh | shOpen the menu
Run `ollama` in a terminal for the interactive menu (official Quickstart).
ollamaRun a model
Model names as listed on ollama.com/library. Local API defaults to http://localhost:11434.
ollama run <model>Pitfalls
Don't start with 70B
Validate on 7B–8B first. Huge weights fill RAM and fans before you've proved the task.
Model licenses ≠ MIT
Ollama's MIT license is not the weights' license. Llama, Qwen, DeepSeek each have terms — check before shipping.
Local API is open by default
Default is http://localhost:11434. Don't publish that port. Keep it on loopback or a private network.
Beside Dify & n8n
Based on official docs and licenses. Ollama is where the model runs; Dify is how you build the app; n8n is how you wire the flow.
| Tool | License | Surface | Positioning | Where data sits |
|---|---|---|---|---|
| Ollama | MIT (models separate) | Local runtime + API | Run open models on your machine | Stays local by default |
| Dify | Apache-2.0 + conditions | AI app workspace | RAG / agent app builder | App can self-host; model calls often leave |
| n8n | Fair-code · SUL | Visual workflows | Integrations and scheduled jobs | Depends on how you deploy |
Official Directory
Official
- Official site (ollama.com)
Downloads and product overview.
- GitHub repository
Source. MIT licensed.
Docs & Install
- Documentation
Install, API and integrations.
- Quickstart
Official getting started: download, menu, first chat.
- Model library
Pullable models — names as listed on the library page.
Related
- Download
macOS / Windows / Linux installers.
Latest Updates
2026-09-04 · GitHub · ollama/ollama
Ollama: one command to run open models locallyMIT local model runtime with an OpenAI-compatible API (default http://localhost:11434). 180k+ stars.
LaoJin's Pick
FAQ
- What is Ollama?
- Ollama is a local runtime for open models (MIT). One command to pull and chat; data stays on the machine by default. It's a runner, not a ChatGPT-style brand.
- How do I install Ollama?
- macOS/Linux: `curl -fsSL https://ollama.com/install.sh | sh`. Windows: the official installer or `irm https://ollama.com/install.ps1 | iex`. Docker image `ollama/ollama` also exists. Then run `ollama`.
- Is Ollama free? Can I use models commercially?
- The runtime is MIT and free. Each model has its own license — some allow commercial use, some don't. Check the library page before shipping a product.
- Ollama vs Dify vs cloud APIs?
- On-machine/private inference → Ollama. Visual RAG apps → Dify with Ollama as a backend. Max quality and you accept data leaving → cloud APIs.
Need to deploy a team-level Agent system?
Open-source frameworks are free, but turning a runtime into an enterprise asset requires access control, RBAC sandboxes, and private RAG integrations. Laojin Global provides commercial-grade architecture and deployment support.
Cite This Hub
We encourage you to capture this hub into your knowledge base, research, or feed it straight to an AI agent under CC BY 4.0.
How this hub stays current
Version and links live in src/content/ollama.ts; star snapshots sync via GitHub API.