llama.cpp: unknown model architecture

The message

llama_model_load: error loading model: unknown model architecture: '<arch>'

Parts in <angle brackets> change from run to run.

What it means

Your llama.cpp build does not know this model family yet. New architectures reach llama.cpp first, then Ollama and LM Studio when they update.

Why it happens

• Your llama.cpp (or the app embedding it) is older than the model.
• The architecture has not been added to llama.cpp at all yet — the GGUF came from a fork.

How to fix it

1. Update: pull and rebuild llama.cpp, or install the latest Ollama / LM Studio.
2. Check the name in the error against llama.cpp's src/llama-arch.cpp; if it is not there, no release can load the file yet. This site's pending page lists released models still waiting on runtime support.

Related errors

• llama.cpp: wrong number of tensors
• Ollama: model may be incompatible with your version