“NVIDIA's TensorRT LLM provided day-zero support for Google's Gemma 4 model.”