Digital Event Horizon
NVIDIA Unveils Nemotron 3.5 Super VL and Bolt Models, Enhancing Efficiency and Customizability in AI Agent Deployments. The company has announced the expansion of its Nemotron family with the introduction of Nemotron 3.5 Super VL and Nemotron 3.5 Bolt models, designed to deliver unparalleled efficiency and customizability in AI agent deployments.
NVIDIA introduces Nemotron 3.5 Super VL and Nemotron 3.5 Bolt models for AI agent deployments. Nematron 3.5 Super VL delivers leading accuracy on multimodal agentic tasks and achieves a 5x faster throughput. Nematron 3.5 Bolt balances intelligence and speed per GB, suitable for local agentic workloads. The models are available now for Nematron 3.5 Super VL, and October 23 for Nematron 3.5 Bolt E4B 4GB. NVIDIA releases open datasets on Hugging Face for agentic tool use, finance, and vision-language understanding. The models run on DGX Spark, DGX Station, and RTX systems for local intelligence, and across data center and cloud environments. The Nematron 3.5 Super VL model is designed for a wide range of applications, including optical character recognition and understanding images and webpages. The model is open for maximum control, security, and privacy, with customization supported through NVIDIA NeMo and open datasets.
NVIDIA has announced the expansion of its Nemotron family with the introduction of Nemotron 3.5 Super VL and Nemotron 3.5 Bolt models. These new models are designed to deliver unparalleled efficiency and customizability in AI agent deployments, catering to a wide range of applications and industries.
The Nemotron 3.5 Super VL model is built in collaboration with the Nemotron Coalition and delivers leading accuracy on multimodal agentic tasks, including instruction following, tool use, multistep business workflows, and visual understanding. This model is designed to handle high-volume agentic workflows and achieves a 5x faster throughput compared to other models in its class.
On the other hand, Nemotron 3.5 Bolt 74B-A9B 24GB is designed to balance intelligence and speed per GB, making it suitable for local agentic workloads. This model is optimized for everyday agentic tasks and can run alongside other applications on NVIDIA RTX/DGX Spark systems. Nemotron 3.5 Bolt E4B 4GB is also available, optimized for intelligence per GB and suitable for fast conversational agents and physical and conversational agents on GeForce RTX, Jetson, and RTX/DGX Spark systems.
The Nemotron 3.5 Super VL model is available now, while Nemotron 3.5 Bolt 74B-A9B 24GB is available now, with Nemotron 3.5 Bolt E4B 4GB coming on October 23. These models are part of the Nemotron family, which also includes Lightning and Ultra.
Nemotron 3.5 Super VL was trained on about 26 trillion tokens across pretraining and continued pretraining, then post-trained with roughly 3.8 million samples and reinforcement learning across dozens of environments. NVIDIA is also releasing open datasets on Hugging Face spanning agentic tool use, finance, instruction following, long context, fairness and vision-language understanding, along with the reinforcement learning environments in NeMo Gym.
The Nemotron 3.5 Super VL model is available now on Hugging Face, Open Router, ModelScope community, and as NVIDIA NIM. It runs on DGX Spark, DGX Station, and RTX systems for local intelligence at the desktop, on NVIDIA Jetson at the edge, and across data center and cloud environments.
Nemotron 3.5 Bolt 74B-A9B 24G is available now on Hugging Face, with support for llama.cpp, Ollama, LM Studio, Unsloth, and vLLM. Built for NVIDIA RTX Spark laptops arriving October 16, it also runs on GeForce RTX PCs, DGX Spark, DGX Station, and RTX PRO workstations.
NVIDIA is also releasing open datasets on Hugging Face spanning agentic tool use, finance, instruction following, long context, fairness and vision-language understanding, along with the reinforcement learning environments in NeMo Gym.
NVIDIA CEO Jensen Huang will deliver a keynote at GTC Berlin on October 21, 11 a.m.-1 p.m. CEST.
In addition, dozens of organizations across industries are post-training Super VL on their own data, tools, policies and workflows to push accuracy higher on the specialized tasks that define their agent deployments. NVIDIA is also announcing new investments in the UK startup ecosystem.
The Nemotron 3.5 Super VL model is designed to handle a wide range of applications, including optical character recognition, tables, charts, and structured information in visual documents. It processes a web page, extracts text, reads graphs and charts, and returns insights from what it finds. This model is also designed to understand images, webpages, and PDFs and act on that understanding.
The Nemotron 3.5 Super VL model is open for maximum control, security, and privacy. Organizations can run the model wherever their security and data-residency requirements dictate, and they also choose to adopt a model they can adapt, rather than one subject to a provider's restrictions. Customization is supported through NVIDIA NeMo, Python-based pipelines, and open datasets.
The Nemotron 3.5 Super VL model is a significant expansion of NVIDIA's AI offerings, providing unparalleled efficiency and customizability in AI agent deployments. With its ability to handle high-volume agentic workflows, achieve a 5x faster throughput, and provide a wide range of applications, this model is poised to revolutionize the AI landscape.
Related Information:
https://www.digitaleventhorizon.com/articles/NVIDIA-Unveils-Nemotron-35-Super-VL-and-Bolt-Models-Enhancing-Efficiency-and-Customizability-in-AI-Agent-Deployments-deh.shtml
https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/
Published: Thu Oct 15 12:50:24 2026 by llama3.2 3B Q4_K_M
