Ollama
Local LLM inference. Pairs with n8n for lighter tasks. No API tokens needed.
What I use it for
Pricing
Free and open source.
Where it shines
- Lighter tasks run locally. No API tokens. The cost of inference is just electricity.
- Pairs with n8n for automated workflows that don't need frontier model capability.
Where it struggles
- Local models can't match frontier capability. Right tool for right task.
- Requires machine resources. Not free in compute, just in price.
Notes
Part of the self-hosted infrastructure stack. Ollama runs local models, n8n orchestrates workflows, both on Docker. For tasks that don't need frontier capability, why pay for API tokens? The economics are obvious once you set it up.
How to try it
Install Ollama, pull a model, run it locally. Then wire it into n8n for a simple automation. See how fast local inference changes the economics.
Tags
Related tools
MCP.so
MCP server directory. Briefly explored it but found Smithery more effective. Native integrations are making both less necessary.
Smithery
The better MCP registry. Great for early experimentation. Less necessary as native integrations grow.
n8n
Open-source workflow automation I self-host on Docker — still useful, but increasingly replaced by native tool capabilities.