NVIDIA Nemotron Nano V2 VL
Abstract
We introduce Nemotron Nano V2 VL, the latest model of the Nemotron vision-language series designed for strong real-world document understanding, long video comprehension, and reasoning tasks. Nemotron Nano V2 VL delivers significant improvements over our previous model, Llama-3.1-Nemotron-Nano-VL-8B, across all vision and text domains through major enhancements in model architecture, datasets, and training recipes. Nemotron Nano V2 VL builds on Nemotron Nano V2, a hybrid Mamba-Transformer LLM, and innovative token reduction techniques to achieve higher inference throughput in long document and video scenarios. We are releasing model checkpoints in BF16, FP8, and FP4 formats and sharing large parts of our datasets, recipes and training code.
Cite
@article{arxiv.2511.03929,
title = {NVIDIA Nemotron Nano V2 VL},
author = {NVIDIA and : and Amala Sanjay Deshmukh and Kateryna Chumachenko and Tuomas Rintamaki and Matthieu Le and Tyler Poon and Danial Mohseni Taheri and Ilia Karmanov and Guilin Liu and Jarno Seppanen and Guo Chen and Karan Sapra and Zhiding Yu and Adi Renduchintala and Charles Wang and Peter Jin and Arushi Goel and Mike Ranzinger and Lukas Voegtle and Philipp Fischer and Timo Roman and Wei Ping and Boxin Wang and Zhuolin Yang and Nayeon Lee and Shaokun Zhang and Fuxiao Liu and Zhiqi Li and Di Zhang and Greg Heinrich and Hongxu Yin and Song Han and Pavlo Molchanov and Parth Mannan and Yao Xu and Jane Polak Scowcroft and Tom Balough and Subhashree Radhakrishnan and Paris Zhang and Sean Cha and Ratnesh Kumar and Zaid Pervaiz Bhat and Jian Zhang and Darragh Hanley and Pritam Biswas and Jesse Oliver and Kevin Vasques and Roger Waleffe and Duncan Riach and Oluwatobi Olabiyi and Ameya Sunil Mahabaleshwarkar and Bilal Kartal and Pritam Gundecha and Khanh Nguyen and Alexandre Milesi and Eugene Khvedchenia and Ran Zilberstein and Ofri Masad and Natan Bagrov and Nave Assaf and Tomer Asida and Daniel Afrimi and Amit Zuker and Netanel Haber and Zhiyu Cheng and Jingyu Xin and Di Wu and Nik Spirin and Maryam Moosaei and Roman Ageev and Vanshil Atul Shah and Yuting Wu and Daniel Korzekwa and Unnikrishnan Kizhakkemadam Sreekumar and Wanli Jiang and Padmavathy Subramanian and Alejandra Rico and Sandip Bhaskar and Saeid Motiian and Kedi Wu and Annie Surla and Chia-Chih Chen and Hayden Wolff and Matthew Feinberg and Melissa Corpuz and Marek Wawrzos and Eileen Long and Aastha Jhunjhunwala and Paul Hendricks and Farzan Memarian and Benika Hall and Xin-Yu Wang and David Mosallanezhad and Soumye Singhal and Luis Vega and Katherine Cheung and Krzysztof Pawelec and Michael Evans and Katherine Luna and Jie Lou and Erick Galinkin and Akshay Hazare and Kaustubh Purandare and Ann Guan and Anna Warno and Chen Cui and Yoshi Suhara and Shibani Likhite and Seph Mard and Meredith Price and Laya Sleiman and Saori Kaji and Udi Karpas and Kari Briski and Joey Conway and Michael Lightstone and Jan Kautz and Mohammad Shoeybi and Mostofa Patwary and Jonathen Cohen and Oleksii Kuchaiev and Andrew Tao and Bryan Catanzaro},
journal= {arXiv preprint arXiv:2511.03929},
year = {2025}
}