Properly local too with the llama and onnx format models available! Awesome

I assume I could just adjust the toml to point to deep seek API locally hosted right?