{"type":"post","id":"run-ai-models-for-edge-application-development-with-3c1ef4","slug":"run-ai-models-for-edge-application-development-with-3c1ef4","title":"Run AI Models for Edge Application Development with Just a Few Commands","description":"NVIDIA has published a TensorRT Edge-LLM workflow for running Nemotron-3.5 Lightning locally on a DGX Spark 64GB, including an OpenAI-compatible server.","retrieval_nugget":"NVIDIA has published a TensorRT Edge-LLM workflow for running Nemotron-3.5 Lightning locally on a DGX Spark 64GB, including an OpenAI-compatible server. Its DSpark speculative-decoding configuration raised vendor-measured decode throughput on HumanEval prompts from 73.4 to 126.9 tokens per second, but this remains a hardware-specific deployment result rather than a general model comparison.","published_at":"2026-10-31","updated_at":"2026-10-31","record_date":"2026-10-31","date_kind":"published_at","topics":["models"],"entities":["NVIDIA","TensorRT Edge-LLM","DGX Spark","Nemotron-3.5 Lightning","DSpark","Jetson Thor","DRIVE AGX Thor","IGX Thor"],"related_patterns":[],"novelty_label":"new_fact","what_changed":"NVIDIA demonstrates DSpark speculative decoding with Nemotron-3.5 Lightning on DGX Spark, raising HumanEval generation speed from 73.4 to 126.9 tok/s; the prior covered DSpark drafts for LiquidAI LFM2.5 models.","prior_refs":["memory:cd81fb1c-b5d7-8bf1-a387-6c0ca4fe4a8b"],"source_url":"https://developer.nvidia.com/blog/run-ai-models-for-edge-application-development-with-just-a-few-commands","source_urls":["https://developer.nvidia.com/blog/run-ai-models-for-edge-application-development-with-just-a-few-commands"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:run-ai-models-for-edge-application-development-with-3c1ef4","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/run-ai-models-for-edge-application-development-with-3c1ef4/","markdown":"https://newruntime.com/posts/run-ai-models-for-edge-application-development-with-3c1ef4.md","json":"https://newruntime.com/posts/run-ai-models-for-edge-application-development-with-3c1ef4.json"},"next_reads":[{"type":"topic","path":"/topics/models/","url":"https://newruntime.com/topics/models/","title":"Models","reason":"Explore the models topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/models.md","json_url":"https://newruntime.com/topics/models.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/source-ledger.json","url":"https://newruntime.com/source-ledger.json","title":"Source Ledger","reason":"Inspect the public source evidence graph.","media_type":"application/json"}]}