Comment by phegler

Okay, this explains the feeling that I had that vLLMs tool parsing is much more precise than for example llama cpp or Ollama, I still use ollama due to mlx which of my understanding is not offered on vLLM unfortunately.