Is NVIDIA TensorRT still maintained, or is it a dead project?

Despite the often fleeting attention given to many open-source projects, NVIDIA's TensorRT is decidedly active, not just ticking along but thriving with regular updates and significant ongoing support. It’s a strong reminder that some legacy frameworks are far from dead.

Current Status of TensorRT

1. TensorRT Core

   • Repository: NVIDIA/TensorRT
   • Stars: 13,108
   • Last Pushed: 2 days ago (June 25, 2026)
   • Maintenance: Actively maintained
   • Open Issues: 608
   • License: Apache-2.0

   TensorRT is NVIDIA's flagship SDK for high-performance deep learning inference on its GPUs. The latest version (11.1) released recently comes with new capabilities aimed at enhancing performance and usability. The active community around TensorRT, with 88 contributors, suggests a robust level of support (for those counting, that’s a healthy bus factor).

2. TensorRT LLM

   • Repository: NVIDIA/TensorRT-LLM
   • Stars: 13,978
   • Last Pushed: 0 days ago (June 27, 2026)
   • Maintenance: Actively maintained
   • Open Issues: 1,431
   • License: NOASSERTION

   This repository is tailored towards optimizing inference for Large Language Models (LLMs). It also has received recent updates, suggesting significant traction and development focus from NVIDIA. With a staggering 411 contributors, its vigor complements the core TensorRT project.

The Verdict

✅ Healthy: Both TensorRT projects are thriving, with implementations pushed often and clearly defined roadmaps. They stand in stark contrast to many other frameworks that may languish or become stale.

Why It Matters

NVIDIA's commitment to TensorRT not only underscores its importance in the AI landscape but also serves as a testament to the advancing capabilities of hardware-accelerated deep learning. If you're considering integrating a high-performance inference solution, TensorRT is not just supported — it is actively evolving, catering to the next-gen needs of AI applications. Whether for traditional deep learning tasks or cutting-edge LLMs, it remains a solid choice.

In summary: If you need a reliable, cutting-edge tool for deep learning inference, look no further than TensorRT; it's firmly on the map and marching forward.