AITOPAI热报
ModelsMulti-source

英伟达开源 Nemotron 3.5 Super VL：120B 参数原生视觉多模态模型

Gorden Sun@Gorden_Sun
Post· 11 sources
View the post
Why we picked it

英伟达又开源了，这次是 120B 总参、12B 激活的视觉模型，主打推理速度，想跑多模态的可以去试试。

英伟达开源了 Nemotron 3.5 Super VL，一个原生视觉多模态模型。

Image · Gorden Sun
Body

模型总参数 120B，激活参数 12B。整体能力属于中规中矩水平，特色在于专门的计算优化，推理速度比同类模型快不少。

模型权重已在 Hugging Face 开放下载。

Original · Gorden Sun

英伟达开源Nemotron 3.5 Super VL 原生视觉多模态模型，120B总参数，12B激活参数，能力中规中矩，特色是有专门的计算优化，速度比同类模型快的多。

模型：https://t.co/8GM7xKqGWS

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Nemotron, NVIDIA, 多模态, 开源模型.

Source attribution: curated from Gorden Sun (@Gorden_Sun). Cross-source corroboration: 11 independent sources have covered this development (including NVIDIA AI, TestingCatalog, Thomas Wolf), confirming industry-wide signal strength.

#Nemotron#NVIDIA#多模态#开源模型

Keep reading

Recent stories with the same tag or section
ProductsBuildersMulti-sourceSelected03:03
CoreWeave webinar on hot-swapping RL checkpoints in production

The non-obvious point is that RL cadence, not compute size, becomes the serving bottleneck once weights update every few hours, and hot-loading removes the downtime tradeoff. ML platform engineers and inference teams running post-training pipelines should attend to copy the Dynamo and RL Rollouts setup directly, including cost figures. Comparing on-policy weight transition approaches before committing to an orchestration stack could save a costly rework later.

#NVIDIA#Nemotron#CoreWeave#Dynamo
Event
CoreWeave@CoreWeaveSource
ProductsBuildersMulti-source12:15
Chinese contentNVIDIA 与 ElevenLabs 扩大合作，Nemotron 模型接入 ElevenAgents 语音智能体

NVIDIA 把自家的 Nemotron 模型塞进了 ElevenLabs 的 ElevenAgents 语音智能体平台，企业和政府现在可以一站式搭建语音客服、语音助手这类应用。做企业级语音方案或政府采购相关项目的开发者可以看看集成方式和部署选项。

#NVIDIA#ElevenLabs#Nemotron#语音智能体
Event
Official accountNVIDIA@nvidiaSource
ModelsBuildersMulti-source02:14
Chinese contentNVIDIA Nemotron Labs 专家直播：详解 Dory Looped Transformer 模型

NVIDIA 的 Nemotron Labs 开了一场专家直播，专门讲 Dory 这个 Looped Transformer 模型的架构设计。想搞清楚循环式 Transformer 跟标准堆叠式架构差在哪、参数量怎么省的朋友，可以直接看回放。做模型结构研究或推理优化的人会有收获。

#Nemotron#NVIDIA#Looped Transformer#推理模型
Event
Official accountNVIDIA AI@NVIDIAAISource
ProductsBuildersMulti-sourceSelected01:20
NVIDIA releases Nemotron-Personas-Japan V2 and PII Japan datasets

Fully synthetic data sidesteps the privacy-approval cycle that typically stalls Japanese-language projects, letting teams start training immediately. Model builders working on PII redaction for regulated sectors like finance and healthcare get ready-made annotations instead of hand-labelling documents. The 10-million-persona scale is rare among Japanese-language datasets, which matters for teams fine-tuning models for the Japan market rather than relying on English-centric synthetic personas.

#NVIDIA#Nemotron#synthetic data#PII detection
Event
Official accountNVIDIA AI@NVIDIAAISource