NVIDIA released Nemotron 3.5 Super VL on October 16, saying it delivers the fastest task completion and highest accuracy for specialized always-on agents. It is the company's first major agentic model update since the release of Nemotron Lightning in August.
NVIDIA reported 5x faster throughput on the Artificial Analysis Intelligence Index leadboard, measured against other models in its class. That compares with the performance of NVIDIA’s largest reasoning model, which it claims is roughly five times larger.
Super VL is built on NVIDIA’s full hardware stack and targets high-volume agentic workflows. Availability begins now, initially for large enterprise servers and edge devices.
"Super VL delivers frontier-level intelligence in its class for high-volume agentic workflows," said NVIDIA. "It helps enterprise agents complete high-volume tasks faster across workflows that span days or weeks."
The announcement follows the release of Bolt models, which are optimized for edge deployment. NVIDIA framed the significance of Super VL as a step toward full coverage of agent deployments, from large servers down to local hardware.
NVIDIA did not say how the model will perform on visual tasks beyond what it described, and raised the question of how well it will handle real-world scenarios. The company said organizations are already post-training Super VL on their own data for specialized tasks.
Source: nvidia
