Getting Started

Getting Started has one goal: run one Qwen request through TensorRT-Model-Connect. Follow these pages in order.

Step	Page	You are done when
1	System Requirements	The GPU is visible and you selected wheel or source.
2	Installation	The installed CLI reports TensorRT support.
3	Build from Source	Optional: the CLI and native DSOs are built for the selected GPU.
4	Quick Start	./qwen3-0.6b.bundle is built, inspected, and returns generated text.

The Quick Start uses a bounded cache profile intended for the first portable inference. A successful run proves the selected environment can resolve the checkpoint, build a bundle, load the native Qwen runtime, and execute one text request. It does not qualify every model or hardware profile on that machine.

After the first run, continue to Learning Path.