# NVIDIA Releases Nemotron 3.5 Super VL, a 120B Open Multimodal Model, and Bolt Edge Models

Times of AI Desk · 2026-10-16 · Models

[https://timesof.ai/2026/10/nvidia-nemotron-3-5-super-vl-bolt](https://timesof.ai/2026/10/nvidia-nemotron-3-5-super-vl-bolt)

> NVIDIA released Nemotron 3.5 Super VL on Oct. 15, an open 120B-parameter multimodal model for always-on agents, with open weights, datasets and recipes, plus Nemotron 3.5 Bolt 74B-A9B for local use in 24GB. Speed and accuracy claims are NVIDIA's.

**NVIDIA** on **Thursday, October 15** released **Nemotron 3.5 Super VL**, an open 120-billion-parameter model for always-on enterprise agents that read web pages, PDFs, charts and documents. NVIDIA calls it "the fastest open agentic model in its class — 5x faster than the next best." That is NVIDIA's claim.

The model ships with open weights, open datasets and training recipes and was trained on about 26 trillion tokens, NVIDIA says. It also released **Nemotron 3.5 Bolt 74B-A9B**, which uses native 2-bit quantization to fit in 24GB for local agents, with a 4GB **Bolt E4B** due October 23. NVIDIA names organizations customizing or integrating Super VL, including Scale AI, CrowdStrike, Perplexity, Salesforce, Atlassian and Ramp.

The models join the family that began with [Nemotron 3.5 Lightning](https://timesof.ai/2026/08/nvidia-nemotron-3-5-lightning-switchyard) in August.

## Limits

- Speed, accuracy and adoption claims are NVIDIA's; the licence text was not read.

## Sources

- [NVIDIA blog: NVIDIA Expands Nemotron Family With Super VL and Bolt Models (October 15, 2026)](https://blogs.nvidia.com/blog/nemotron-3-5-super-vl/)
- [Times of AI: Nemotron 3.5 Lightning and Switchyard (August 11, 2026)](https://timesof.ai/2026/08/nvidia-nemotron-3-5-lightning-switchyard)