NVIDIA released Nemotron 3.5 Super VL on October 15, 2026. It is an open 120-billion-parameter vision-language model built for always-on AI agents. NVIDIA also announced two smaller edge models, Nemotron 3.5 Bolt 74B-A9B 24GB and Nemotron 3.5 Bolt E4B 4GB.
Super VL and Bolt 74B-A9B are available now. Bolt E4B is coming October 23.
Three models, three tiers
- Nemotron 3.5 Super VL: 120B parameters, for enterprise agents running high-volume workflows.
- Nemotron 3.5 Bolt 74B-A9B 24GB: for everyday agentic tasks on local hardware.
- Nemotron 3.5 Bolt E4B 4GB: for fast conversational and physical agents.
They join the existing Nemotron Lightning (released in August) and Ultra models.
Super VL: 5x faster than the next best in its class
NVIDIA says Super VL is the fastest open agentic model in its class, 5x faster than the next best, based on the Artificial Analysis Intelligence Index. NVIDIA says it reaches accuracy comparable to its largest reasoning model at roughly one-fifth the parameter count, on tasks such as instruction following, tool use, multistep business workflows and visual understanding.
Its vision skills target what agents see in practice: OCR, tables, charts and structured information in documents. It can process web pages and PDFs, extract text and read graphs.
Fully open
The model ships with open weights, open datasets and training recipes. NVIDIA says it was trained on about 26 trillion tokens, then post-trained with roughly 3.8 million samples and reinforcement learning across dozens of environments. NVIDIA is also releasing datasets on Hugging Face and reinforcement learning environments in NeMo Gym.
Super VL is available on Hugging Face, OpenRouter, the ModelScope community and as NVIDIA NIM. NVIDIA says that running on NVIDIA Dynamo lets organizations serve up to 8x more users with 85% lower latency.
Bolt: agents on your own machine
Bolt 74B-A9B 24GB uses native 2-bit quantization to fit in 24GB of memory, so it can run on NVIDIA RTX and DGX Spark systems alongside other applications. NVIDIA says it is the #1 open model in its class on PinchBench and the #2 open model on BrowseComp. It is on Hugging Face now, with support for llama.cpp, Ollama, LM Studio, Unsloth and vLLM. It is built for NVIDIA RTX Spark laptops, which arrive October 16.
Bolt E4B 4GB targets voice assistants, interactive characters and embedded interfaces on GeForce RTX, Jetson and RTX/DGX Spark.
Who is using it
NVIDIA names Atlassian, CrowdStrike, Dassault Systèmes, Perplexity, Salesforce and Synopsys among the software companies integrating Super VL. Scale AI, Mercor, AI21, Ramp and others are post-training it for specialized tasks.
