NVIDIA Releases TensorRT Model Connect in Public Preview: Hugging Face Checkpoint to Native C++ Inference in Two Commands
Two commands: from Hugging Face to production C++ inference. NVIDIA's TensorRT Model Connect eliminates the ONNX middleman.
Why it matters
TensorRT Model Connect streamlines the model-to-deployment pipeline for practitioners building inference services, removing friction from checkpoint to optimized native code — a meaningful quality-of-life improvement for production ML teams.
The key facts
7 to knowTensorRT Model Connect released in public preview
Apache 2.0 license
Converts Hugging Face or local checkpoints to TensorRT inference in two commands
No ONNX export step required
Outputs versioned .bundle artifacts running through native C++ task APIs
Runtime executes without PyTorch in the path
105 release profiles across 76 model families supported (July 29, 2026 GB300 snapshot)
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: NVIDIA has released TensorRT Model Connect (TRTMC) in public preview, an Apache-2.0 project that takes a supported Hugging Face or local checkpoint to end-to-end TensorRT inference in two commands, with no intermediate ONNX export. The build emits a versioned .bundle artifact that runs through…