During the last AI hype we saw some packages for using the OpenAI API to work with dynamo Revit. I lost interest due to the paywall, yet i had a few ideas i wanted to fiddle with. Does anyone in the forum have the programing know how to use the open source LLM in a dynamo workflow. let me know if i may have missed any packages.
Welcome to the forum.
Pretty sure there was a post about this quite recently. Try searching the forum. 
To package up deepseek would require packaging its data sets as well (which are huge). The ‘open data’ (not open source) nature of deepseek is you can set it all up locally, but this means doing a lot of setup i cant see a dynamo package achieving.
That and you will need a pretty high end setup with graphics cards etc to my understanding based on the guides I have checked.
Suggest you google some examples of setup so you can see what I mean. Most of them involve some setup related to llama etc.
Thanks for the fast response. I have seen a few version of Deepseek with low memory like 32gb version with ollama, but not sure how good it would work. Preferably doing it locally would be great for Security purposes, but it would be great to test out similar to OpenAI without jumping into a local version to start.
The web/app versions are not open source, but have a web api.
Tokens are cheaper but not free:
Webapi:
