Those are not deepseek R1. They are unrelated models like llama3 from Meta or Qwen from Alibaba “distilled” by deepseek.

This is a common method to smarten a smaller model from a larger one.

Ollama should have never labelled them deepseek:8B/32B. Way too many people misunderstood that.

The hottest model now is flux, and the community is gravitating toward it after the failure of stable diffusion 3.

Tooling around it is still nascent and it is very demanding (12GB+ VRAM GPU highly recommended) so you could also start with Stable Diffusion XL, fairly mature at this point.

So far seem to be an uneventful upgrade.

Defaulting to wayland for KDE6 on a nvidia GPU doesn’t seem to have broken anything