There's a few possible answers for direction but...
The architecture that is shown on the Foundation screen "Overview" section I believe is getting pulled from config.json for the model. However we store whatever is in the gallery as architecture and override that. So for example, with mlx-community/Llama-3.2-1B-Instruct-4bit we are using MLX internally as the architecture but the screen shows this
In this case, that's just a bit of inconsistent info...but it has caused me debugging hell because I couldn't figure out why some models were and weren't showing as having available plugins!