NVIDIA released Nemotron 3.5 Super VL and a 24GB Bolt model, while a smaller 4GB Bolt conversational model is scheduled for October 23.
NVIDIA announced Nemotron 3.5 Super VL on October 15, 2026, describing it as an open, 120-billion-parameter model for multimodal agent tasks. The company says it is designed to handle instruction following, tool use, business workflows and visual understanding, including reading information in documents, webpages, charts and tables. NVIDIA says the model completes tasks five times faster than the next-best model in its class, and reports accuracy comparable to its largest reasoning model at roughly one-fifth the parameter count.
The release also adds two models aimed at local deployment. Nemotron 3.5 Bolt 74B-A9B 24GB uses native 2-bit quantization to fit within 24GB of memory and is intended for agent workloads on NVIDIA RTX and DGX Spark systems. NVIDIA says it is available now on Hugging Face and supports tools including llama.cpp, Ollama, LM Studio, Unsloth and vLLM. Nemotron 3.5 Bolt E4B 4GB is designed for conversational and physical agents on GeForce RTX, Jetson and RTX/DGX Spark hardware; NVIDIA says it is expected on October 23.
Super VL is available now through Hugging Face, OpenRouter, ModelScope and NVIDIA NIM, and is designed to run across NVIDIA data-center, workstation and edge hardware. NVIDIA says organizations can adapt it using their own data, tools and workflows. The release includes open weights, datasets and training recipes; customization is supported through NVIDIA NeMo and Python-based pipelines.
NVIDIA reports that Super VL was trained on about 26 trillion tokens and post-trained with roughly 3.8 million samples, alongside reinforcement learning across dozens of environments. The company lists applications including finance-document analysis, cybersecurity, legal research and coding. It also names organizations customizing or integrating the model, among them Atlassian, CrowdStrike, Perplexity and Salesforce. These are company-reported examples, rather than independent evaluations of the models’ performance.
The three releases extend NVIDIA’s Nemotron lineup from data-center deployments toward local systems. Super VL and Bolt 74B-A9B are available now; the smaller Bolt model has a stated October 23 release date.