Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.33039

Nemotron 3.5 Bolt 74B-A9B 24GB: A Hybrid Vision-Language Model for Local AI

Ganesh Ajjanagadde·Chris Alexiuk·Luke Alonso·Kumar Anshuman·Maor Ashkenazi·Nave Assaf·Gorkem Batmaz·Omer Belhasin·Shai Ben Nun·Akhiad Bercovich·Pritam Biswas·Caroline Blanco·Tijmen Blankevoort·Michael Boone·Christopher Borden·Kari Briski·Bryan Catanzaro·Zijia Chen·Aishwarya Chhabra·Hung-Yueh Chiang·Annamalai Chockalingam·Jonathan Cohen·Joey Conway·Olivier Delalleau·Gerardo Delgado Cabrera·Leon Derczynski·Li Ding·Will Efstratis·Michael Elad·Ellie Evans·Wes Feely·Michael Fukuyama·Gaurav Garg·Rishabh Garg·Arvind Gopalakrishnan·Shishir Goyal·Kartik Gulia·Hao Guo·Negar Habibi·Scott Hara·Ehsan Hosseini Asl·Rohit Jena·Udi Karpas·Bilal Kartal·Mayank Kaushik·Manoj Kilaru·Mackenzie Kosseim·Oleksii Kuchaiev·Matthieu Le·Raymond Lee·Jian Leong·Jonathan Li·Sheldon Liang·Mingjie Liu·Eileen Long·Katherine Luna·Chenjie Luo·Mingyuan Ma·Arsh Maknojia·Richard Mazzarese·Jessica Medina·Puhui Meng·Yoav Miron·Pavlo Molchanov·David Mosallanezhad·Chetan Mungekar·Marina Neseem·Oluwatobi Olabiyi·Jason Paul·Bhavesh Pawar·Seth Poulos·Anav Prasad·Meredith Price·Sweta Priyadarshi·Haifeng Qian·Vinay Raman·Kaustubh Rao·Hayley Ross·Matan Rusanovsky·Sudeep Sabnis·Karan Sapra·Micah Schaffer·Schmitty·Alex Settle·Ashton Sharabiani·Haran Kumar Shiv Kumar·Rameshwar Shivbhakta·Guillermo Siman·Barnaby Simkin·Oliver Simons·Andrii Skliar·Padmavathy Subramanian·Yoshi Suhara·Ali Taghibakhshi·Nima Tajbakhsh·Andrew Tao·Roy Uziel·Prasoon Varshney·Jose Vergara·Huy Vu
View original source ↗Read source document ↗Export metadata ↓

Abstract

We introduce Nemotron 3.5 Bolt 74B-A9B 24GB, a hybrid Mamba-Transformer vision-language model that optimizes speed and intelligence per GB of memory used for local agentic workloads. Derived from the Nemotron 3.5 Super VL 120B-A12B model family, Nemotron 3.5 Bolt 74B-A9B 24GB combines Star Elastic compression with 2-bit Quantization-Aware Distillation for memory-constrained deployment. The model is optimized for reasoning, agentic tool use, and web & terminal interactions. We also trained a LiLiCorr speculative drafter that is released alongside the model to reach our target inference speed. The resulting model achieves a higher accuracy than leading models with similar memory footprints and speed across many benchmarks for general agentic capabilities (e.g., GDPVal v2), web browsing (e.g., BrowseComp), and terminal use (e.g., Terminal-Bench 2.1), with a relatively high TPS of 84 on RTX Spark that is comparable to models with much lower active parameter counts and more than sufficient for long-horizon agentic applications.

Explore related subjects

cs.CV

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ganesh Ajjanagadde, Chris Alexiuk, Luke Alonso, Kumar Anshuman, Maor Ashkenazi, Nave Assaf, Gorkem Batmaz, Omer Belhasin, Shai Ben Nun, Akhiad Bercovich, Pritam Biswas, Caroline Blanco, Tijmen Blankevoort, Michael Boone, Christopher Borden, Kari Briski, Bryan Catanzaro, Zijia Chen, Aishwarya Chhabra, Hung-Yueh Chiang, Annamalai Chockalingam, Jonathan Cohen, Joey Conway, Olivier Delalleau, Gerardo Delgado Cabrera, Leon Derczynski, Li Ding, Will Efstratis, Michael Elad, Ellie Evans, Wes Feely, Michael Fukuyama, Gaurav Garg, Rishabh Garg, Arvind Gopalakrishnan, Shishir Goyal, Kartik Gulia, Hao Guo, Negar Habibi, Scott Hara, Ehsan Hosseini Asl, Rohit Jena, Udi Karpas, Bilal Kartal, Mayank Kaushik, Manoj Kilaru, Mackenzie Kosseim, Oleksii Kuchaiev, Matthieu Le, Raymond Lee, Jian Leong, Jonathan Li, Sheldon Liang, Mingjie Liu, Eileen Long, Katherine Luna, Chenjie Luo, Mingyuan Ma, Arsh Maknojia, Richard Mazzarese, Jessica Medina, Puhui Meng, Yoav Miron, Pavlo Molchanov, David Mosallanezhad, Chetan Mungekar, Marina Neseem, Oluwatobi Olabiyi, Jason Paul, Bhavesh Pawar, Seth Poulos, Anav Prasad, Meredith Price, Sweta Priyadarshi, Haifeng Qian, Vinay Raman, Kaustubh Rao, Hayley Ross, Matan Rusanovsky, Sudeep Sabnis, Karan Sapra, Micah Schaffer, Schmitty, Alex Settle, Ashton Sharabiani, Haran Kumar Shiv Kumar, Rameshwar Shivbhakta, Guillermo Siman, Barnaby Simkin, Oliver Simons, Andrii Skliar, Padmavathy Subramanian, Yoshi Suhara, Ali Taghibakhshi, Nima Tajbakhsh, Andrew Tao, Roy Uziel, Prasoon Varshney, Jose Vergara, Huy Vu. 2026-10-15. Nemotron 3.5 Bolt 74B-A9B 24GB: A Hybrid Vision-Language Model for Local AI. https://arxiv.org/abs/2610.33039

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Graph-Aggregated Prototype Learning for Remote Sensing Semantic Change Detection

Semantic change detection (SCD) extends the binary change detection task to provide not only the change locations but also the detailed "from-to" categories in multi-temporal remote sensing data. Such detailed semantic insights into changes offer considerable advantages for a wide array of applications. However, since SCD involves the simultaneous optimization of multiple tasks, the model is prone to negative transfer due to task-specific learning difficulties and conflicting gradient flows. To address this issue, we propose Graph-Aggregated Prototype Learning for Remote Sensing Semantic Change Detection(GAPL-SCD). In this framework, a multi-task joint optimization method is designed to optimize the primary task of semantic segmentation and change detection, along with the auxiliary task of graph aggregation prototype learning. Adaptive weight allocation and gradient rotation methods are used to alleviate the conflict between training tasks and improve multi-task learning capabilities. Specifically, the graph aggregation prototype learning module constructs an interaction graph using high-level features. Prototypes serve as class proxies, enabling category-level domain alignment across time points and reducing interference from irrelevant changes. Additionally, the proposed self-query multi-level feature interaction and bi-temporal feature fusion modules further enhance multi-scale feature representation, improving performance in complex scenes. Experimental results on the SECOND and Landsat-SCD datasets demonstrate that our method achieves state-of-the-art performance, with significant improvements in accuracy and robustness for SCD task.

cs.CV↗
arXiv2026

Compact Attention: Exploiting Structured Spatio-Temporal Sparsity for Fast Video Generation

The quadratic computational complexity of self-attention mechanisms imposes a critical constraint on transformer-based video generation, particularly for synthesizing ultra-long and high-definition sequences. While most sparse attention techniques employ fixed or heuristic patterns, they often struggle to accommodate the heterogeneous and layer-specific redundancies inherent in video diffusion transformers (VDiTs). We find that different attention heads consistently concentrate most of their attention mass in different coarse query-key regions, a behavior we call functional specialization. Although individual attention values vary with the generated content, these regions remain similar across prompts and random seeds. Leveraging these insights, we introduce Compact Attention, a hardware-aware sparse attention framework for efficient long-video synthesis. We quantify this stability and use it to construct searchable, reusable, head-specific masks. Based on the observation, we propose a deformable dual-window sparse pattern mechanism that flexibly approximates local and axial attention regions within one tile-aligned mask family. To avoid manually choosing a mask for each head, we develop a recall-driven offline auto-search algorithm that identifies high-quality sparse masks while leveraging temporal stability for mask reuse. Extensive evaluations on Wan2.1 (14B) and Hunyuan-Video demonstrate that Compact Attention achieves 2.25x/3.03x attention speedups over FlashAttention-2 and 1.20x/1.74x over FlashAttention-3, respectively, while maintaining visual quality and temporal consistency comparable to full-attention baselines. Project Page: https://yo-ava.github.io/Compact-Attention.github.io/

cs.CV↗
arXiv2026

Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment

Test-time adaptation (TTA) enhances the zero-shot robustness under distribution shifts by leveraging unlabeled test data during inference. Despite notable advances, several challenges still limit its broader applicability. First, most methods rely on backpropagation or iterative optimization, which limits scalability and hinders real-time deployment. Second, they lack explicit modeling of class-conditional feature distributions. This modeling is crucial for producing reliable decision boundaries and calibrated predictions, but it remains underexplored due to the lack of both source data and supervision at test time. In this paper, we propose ADAPT, an Advanced Distribution-Aware and backPropagation-free Test-time adaptation method. We reframe TTA as a Gaussian probabilistic inference task by modeling class-conditional likelihoods using gradually updated class means and a shared covariance matrix. This enables closed-form, training-free inference. To correct potential likelihood bias, we introduce lightweight regularization guided by CLIP priors and a historical knowledge bank. ADAPT requires no source data, no gradient updates, and no full access to target data, supporting both online and transductive settings. Extensive experiments across diverse benchmarks demonstrate that our method achieves state-of-the-art performance under a wide range of distribution shifts with superior scalability and robustness.

cs.CV↗