Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models
NVIDIA has released the Nemotron 3.5 Super VL, a 120-billion-parameter model, alongside two edge-optimized models: the 74B-A9B 24GB and the E4B 4GB. These models are designed to support specialized always-on agents and local agentic workloads across data center and desktop environments.
Verified State Diff
Impact & Verification Analysis
AI developers, enterprise infrastructure architects, and edge computing engineers.
It enables the deployment of high-performance, open-weight agentic models on local hardware, significantly lowering latency and operational costs for always-on AI applications.
Full Fact Overview
The expansion of the Nemotron 3.5 family represents a strategic push by NVIDIA to provide scalable AI infrastructure from high-compute data centers to resource-constrained edge devices. The 120B Super VL model targets high-accuracy, high-throughput tasks, while the Bolt series utilizes specific parameter counts (74B and 4B) and memory footprints (24GB and 4GB) to optimize inference speed and local execution. This release signals a shift toward deploying specialized agentic models directly on hardware, reducing latency and dependency on cloud-based API calls.