Comment by stymaar

Is it a fully custom inference engine or are you reusing parts of an existing stack? (llama.CPP, vLLM, etc.)