What is Nemotron 3.5 Lightning and how fast is it?

Nemotron 3.5 Lightning is a customizable open 30-billion-parameter mixture-of-experts model from NVIDIA for always-on agents. It delivers up to 4× faster token generation and 30% faster time to completion compared to open models in its class.

Can these models run on consumer hardware?

Yes. Meta's Muse Glimmer (30-billion-parameter) runs on a PC with a single consumer GPU, delivering over 200 tokens per second on RTX 5090. Nemotron 3.5 Lightning runs locally on NVIDIA RTX PCs, DGX Spark, OEM GB10 systems, and NVIDIA Jetson.

What is Unsloth Desktop?

Unsloth Desktop is a fully open-source desktop app that brings local model inference, image and video diffusion, fine-tuning, agent integrations, web research and code execution into one application. It launched on Monday and is positioned as the first desktop app that both trains and runs AI models locally.

How does NVIDIA Sync Cluster Assistant help?

The Cluster Assistant automates configuring two or more DGX Spark systems as a high-speed cluster. Developers connect systems through NVIDIA ConnectX-7 ports, and NVIDIA Sync configures the network, routes workloads across nodes and monitors system health, enabling larger models to run faster across multiple devices.

Large Language ModelsOpen-Source AITop Companies' AI MovesAI Business & IndustryTop Companies AIAug 11, 2026Aug 12, 2026, 06:30 JST

NVIDIA, partners release wave of open AI models for local inference

Send on LINE

3 Key Points

What happened
NVIDIA and partners released multiple open-source AI models optimized for local execution throughout August, including Meta's Muse Glimmer (30-billion-parameter), DeepSeek-V4-Flash (284-billion-parameter with 1 million-token context), LTX-2.5 (video generation), and NVIDIA's Nemotron 3.5 Lightning (30-billion-parameter mixture-of-experts). NVIDIA also launched new tools: Unsloth Desktop (an open-source app for local training and inference), NVIDIA Sync Cluster Assistant (for linking multiple DGX Spark systems), and upcoming DGX Spark features including native Google Chrome support and a Resource Monitor.
Why it matters
These releases lower the barrier for developers and AI enthusiasts to build and run capable agents entirely on local hardware—PCs, DGX Spark systems, or Jetson devices—without relying on cloud inference. This means faster response times, private data handling, and potentially lower operational costs for use cases like coding agents, local document processing, and multistep task automation. The combination of smaller, optimized models and clustering tools allows a single consumer GPU (like RTX 5090) or linked systems to handle workloads previously requiring cloud resources.
What to watch
Nemotron 3.5 Lightning delivers up to 4× faster token generation and 30% faster time to completion compared to open models in its class. Muse Glimmer processes over 200 tokens per second on RTX 5090. LTX-2.5 achieves up to 20% faster performance and 40% memory savings on an NVIDIA RTX 6000 PRO GPU. New DGX Spark features arrive later in August.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

FAQ

What is Nemotron 3.5 Lightning and how fast is it?: Nemotron 3.5 Lightning is a customizable open 30-billion-parameter mixture-of-experts model from NVIDIA for always-on agents. It delivers up to 4× faster token generation and 30% faster time to completion compared to open models in its class.
Can these models run on consumer hardware?: Yes. Meta's Muse Glimmer (30-billion-parameter) runs on a PC with a single consumer GPU, delivering over 200 tokens per second on RTX 5090. Nemotron 3.5 Lightning runs locally on NVIDIA RTX PCs, DGX Spark, OEM GB10 systems, and NVIDIA Jetson.
What is Unsloth Desktop?: Unsloth Desktop is a fully open-source desktop app that brings local model inference, image and video diffusion, fine-tuning, agent integrations, web research and code execution into one application. It launched on Monday and is positioned as the first desktop app that both trains and runs AI models locally.
How does NVIDIA Sync Cluster Assistant help?: The Cluster Assistant automates configuring two or more DGX Spark systems as a high-speed cluster. Developers connect systems through NVIDIA ConnectX-7 ports, and NVIDIA Sync configures the network, routes workloads across nodes and monitors system health, enabling larger models to run faster across multiple devices.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.