NVIDIA releases Nemotron 3.5 Lightning, an open 30B model with 3B active parameters for agents
Announced · Published · By Contractor AI Shop
NVIDIA released Nemotron 3.5 Lightning, an open mixture-of-experts model with 30B total and 3B active parameters, aimed at fast, repetitive agent steps like tool calls. CNBC reported NVIDIA says it can run on a single GPU in a laptop or desktop. It is available in Ollama.
Key facts
- NVIDIA announced Nemotron 3.5 Lightning on August 11, 2026 as an open 30B mixture-of-experts model with 3B active parameters. NVIDIA
- NVIDIA says it ships BF16 and NVFP4 quantized checkpoints and is built for agent tools such as OpenClaw and Hermes Agent. NVIDIA
- CNBC reported NVIDIA calls the model lightweight and says it can run on a single GPU in a laptop or desktop. CNBC
- OSBBD lists a context window of up to 1M tokens and the OpenMDW License Agreement v1.1, which allows commercial use. OSBBD
- Ollama says you can run it with 'ollama run nemotron-3.5-lightning'. Ollama
Our take: what it means for contractors
With only 3B active parameters, this model is built to be fast on modest hardware, which matters if you want an agent that runs all day on an office PC. Use it for routine steps like sorting emails or filling forms, and keep a stronger model for anything you'd sign your name to.
What we don't know yet
- Measured speed and quality on common small-business hardware
- How well it handles construction documents compared with larger models
Sources
- NVIDIA · NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents · 2026-08-11 · Primary
- CNBC · Nvidia unveils first open-source AI model since CEO Jensen Huang entered the chat · 2026-08-11 · Secondary
- OSBBD · Nemotron 3.5 Lightning: license, size and hardware needs · 2026-09-29 · Secondary
- Ollama · NVIDIA Nemotron 3.5 Lightning · Secondary