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

Computer Science > Computer Vision and Pattern Recognition arXiv:2610.33039 (cs) [Submitted on 15 Oct 2026] Title: Nemotron 3.5 Bolt 74B-A9B 24GB: A Hybrid Vision-Language Model for Local AI.

计算机科学 > 计算机视觉与模式识别 arXiv:2610.33039 (cs) [提交于 2026 年 10 月 15 日] 标题：Nemotron 3.5 Bolt 74B-A9B 24GB：一种用于本地人工智能的混合视觉语言模型。

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.

摘要：我们介绍了 Nemotron 3.5 Bolt 74B-A9B 24GB，这是一种混合 Mamba-Transformer 视觉语言模型，旨在优化本地智能体工作负载中每 GB 内存的运行速度与智能水平。该模型源自 Nemotron 3.5 Super VL 120B-A12B 模型系列，结合了 Star Elastic 压缩技术与 2 位量化感知蒸馏（Quantization-Aware Distillation），以适应内存受限的部署环境。

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.

该模型针对推理、智能体工具使用以及网页和终端交互进行了优化。我们还训练了一个 LiLiCorr 推测草稿器（speculative drafter），并随模型一同发布，以达到预期的推理速度。在通用智能体能力（如 GDPVal v2）、网页浏览（如 BrowseComp）和终端使用（如 Terminal-Bench 2.1）等多个基准测试中，该模型在相似内存占用和速度下，比领先模型实现了更高的准确率。其在 RTX Spark 上达到了 84 的相对高 TPS，这与激活参数量更低的模型相当，且足以满足长程智能体应用的需求。