It would be nice to have a way to update the llama.cpp binaries without having to wait for a new release of OGAD to support the latest models, for example right now the current version of OGAD does not support the new edge models released during july and august 2026 which made huge performance jumps in agentic benchmarks compared to previous generations (eg, Bonsai-27B unary and Nanbeige-4.2-3B). I updated manually the binaries in %USERPROFILE%\AppData\Local\Programs\off-grid-ai\resources\bin\llama and llama-cpp and Nanbeige-4.2-3B from Bartowski runs fine.
I suggest to place it in Settings > System health, it could even be done automatically when the llama.cpp tries to restart in a loop but constantly crashes (it even told me that the server was too old to run this model but I'm not sure this is reliable to expect such a message for all future models).
It would be nice to have a way to update the llama.cpp binaries without having to wait for a new release of OGAD to support the latest models, for example right now the current version of OGAD does not support the new edge models released during july and august 2026 which made huge performance jumps in agentic benchmarks compared to previous generations (eg, Bonsai-27B unary and Nanbeige-4.2-3B). I updated manually the binaries in
%USERPROFILE%\AppData\Local\Programs\off-grid-ai\resources\bin\llamaandllama-cppand Nanbeige-4.2-3B from Bartowski runs fine.I suggest to place it in Settings > System health, it could even be done automatically when the llama.cpp tries to restart in a loop but constantly crashes (it even told me that the server was too old to run this model but I'm not sure this is reliable to expect such a message for all future models).