AI & Tech
NVIDIA Expands Nemotron Family From Data Centers to Desktop and Edge Devices
NVIDIA introduced Nemotron 3.5 Super VL and two Bolt edge models. The Decoder reported that the 24GB memory threshold for Bolt 74B-A9B relates to local-agent use mainly involving developers and high-end workstation users.
AiSiWhale・Published
NVIDIA said the Nemotron family now includes Nemotron 3.5 Super VL and two Bolt edge models, extending the lineup from data centers to desktop and edge devices. NVIDIA positioned Super VL as an open model for specialized persistent agents.
NVIDIA’s technical blog said Super VL delivers up to five times the throughput of other comparable open models for agent tasks on a single GPU. NVIDIA also said Nemotron 3.5 Super VL provides open weights, open datasets and training recipes, allowing enterprises to post-train it with their own data, tools, policies and workflows.
NVIDIA said Bolt 74B-A9B 24GB is aimed at local-agent workloads, while Bolt E4B 4GB is aimed at fast conversational agents. Bolt 74B-A9B uses native 2-bit quantization, can run within 24GB of memory and supports local-agent tasks.
NVIDIA said Super VL and Bolt 74B-A9B are available, while Bolt E4B is expected to launch on October 23, 2026; that year is inferred from the announcement’s publication date of October 15, 2026, because the original quotation says only “coming Oct. 23.”
The Decoder reported that the versions in the Nemotron family are optimized for NVIDIA hardware. The Decoder also said that the 24GB memory threshold for Bolt 74B-A9B relates to local-agent use mainly involving developers and high-end workstation users.
Sources
- Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models | NVIDIA Blog (blogs.nvidia.com)
- NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning | NVIDIA Technical Blog (developer.nvidia.com)
- Nvidia's Nemotron 3.5 Super VL is an open model built to run enterprise AI agents nonstop (the-decoder.com)