NVIDIA releases TensorRT Edge-LLM and DGX Spark 64GB

NVIDIA introduces the TensorRT Edge-LLM runtime and DGX Spark platform, enabling embedded developers to prototype, fine-tune, and run AI models like Nemotron-3.5 Lightning at the edge using simple command-line tools. The platform supports local OpenAI-compatible APIs and DSpark speculative decoding to improve token generation speeds for edge applications. Open article →