-
To elaborate: they have done away with thier GO-Runners for GGUF files completely.
Instead they now favor direct Llama.cpp client/server that they wrap to run GGUF models.
-
Because of this, any Llama3.2-Vision based models are no longer supported with the above mentioned versions or higher.
I however, do have an pull request that I will be submitting this week that adds support for "mllama" model typs (Multimodal-Llama)