// AI news · 2026-10-31 · NVIDIA
NVIDIA Launches TensorRT Edge-LLM for Edge AI Development
NVIDIA introduces TensorRT Edge-LLM to enable AI model development for edge devices on DGX Spark 64GB.
Published 2026-10-31 04:33 (Beijing time)Source NVIDIAAuto-summarised, fact-checked against source
01Summary
The tool provides a lightweight C++ runtime that supports model deployment across Jetson Thor, DRIVE AGX Thor, and IGX Thor platforms. Users can start with Nemotron-3.5 Lightning via a few CLI commands, including model download and engine build. DSpark speculative decoding boosts generation speed to 126.9 tokens per second with peak memory at 27.9GB.
Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.
02Key points
- TensorRT Edge-LLM enables AI development on DGX Spark 64GB for edge devices.
- Models deploy across Jetson Thor, DRIVE AGX Thor, and IGX Thor platforms.
- trtedge CLI downloads and builds optimized engines in a few commands.
- DSpark decoding increases generation speed to 126.9 tokens/sec with 27.9GB peak memory.
03Source
Run AI Models for Edge Application Development with Just a Few Commands
NVIDIA · published 2026-10-31 04:33 Beijing time
05Related news
- Same source · NVIDIA Releases NemoClaw Open Blueprints for Specialized Agents2026-10-31 · NVIDIA
- Same source · NVIDIA Releases gds-diag for GDS Configuration Diagnostics2026-10-30 · NVIDIA
- Same topic · NVIDIA Launches Halloween Interactive Projection with Nemotron2026-10-31 · Hugging Face Blog
- Same topic · Agentic AI Assists in Search for Extraterrestrial Signals2026-10-30 · NVIDIA Newsroom
- Same day · Dartmouth Provost Resigns Over Undisclosed AI Use2026-10-31 · Ars Technica AI
From AIMap AI news: collected from 1,100+ official and media sources. Summaries only restate the original article; follow the link for the full story. All stories of 2026-10-31 · Archive by date · All AI news · Today