Great question! We explored local LLMs (including llamafile-type solutions) in our early development, but found that the reasoning capabilities and consistency weren't quite there yet for our specific needs.
That's why we currently optimize for cloud AI models while implementing intelligent plan caching to significantly reduce API costs. This approach gives you the best of both worlds: high-quality execution plans with minimal API costs, plus much faster performance for similar actions.
You might find our documentation on plan caching interesting - it explains how we maximize efficiency: https://github.com/orra-dev/orra/blob/main/docs/plan-caching...
We're always evaluating new LLM options though, so I'd be curious to hear about your specific use case.
