suggestion: better integration with Llama-server /Llama swap #767
Description
Unfortunately, Ollama is going down the road of the Enshittification. They're trying to promote their models in the cloud, overturning the original vision of being a tool that simplifies local model management.
On the contrary, llama.cpp is constantly improving, adding new features that promote better management of LLMS running locally, providing many interesting metrics and a more detailed customization of the model parameters.
Since llama-server has been improved, the difficulty and inconvenience gap that existed compared to Ollama ease of use has been closed, and coupled with llama swap, having Ollama is becoming increasingly irrelevant.
What I would suggest is allowing page assist to work even without Ollama.
- Currently, when Ollama is missing, a window appears warning that Ollama is missing even though there are llama cpp/openAI models with compatible APIs selected.
- Access the metrics provided by llama.cpp /llama swap
- Currently it is possible to set llama swap as OpenAI api and the list of all importable models in Page assist appears, and this is great, but it would be more interesting if Page assist monitors the change of these models llama swap or the addition of others, currently if the LLMs are manipulated in llama swap, I have to delete the models from page assist and re-import them, it would be useful if this step was done automatically.
