Posit Assistant in Positron allows you to connect any local models from Ollama. I'm curious if anyone has had good results with any of the local models for doing data analysis in R and/or R package development? I've heard gemma 4 and qwen3.6 are supposed to be decent. Curious if anyone has any recommendations for free local models?
1 Like
How much GPU ram do you have in the GPU that you're using with Ollama?
Are you using something like an RTX 4090?
Qwen 3.6 14b is pretty good locally for me with Python for development. 32GB of unified memory. Note that I haven't actually tried this with Posit Assistant yet. I use Qwen 3.6 with custom applications.
1 Like
I’m on a Mac M4 Max (32‑core GPU) with 36 GB RAM. I think the GPU should have access to 25-30GB RAM.
1 Like