NVIDIA推出Nemotron 3 Nano Omni模型，多模态推理效率提升9倍
Odaily星球日报讯 NVIDIA 在 X 平台发文表示，今日推出开源多模态模型 Nemotron 3 Nano Omni。该模型采用 30B-A3B 混合专家架构（MoE），支持 256K 上下文，可统一处理视频、音频、图像及文本输入。与同类交互水平的开源全能模型相比，该模型吞吐量提升达 9 倍，大幅降低推理成本并提高扩展性。目前 Nemotron 3 Nano Omni 已在 Hugging Face、OpenRouter 及 NVIDIA NIM 上线，并获得 Aible、Applied Scientific Intelligence、H Company 等企业采用。
Disclaimer: OKX Orbit content is provided for informational purposes only. Learn more
Replies
Related Flash News
AI startup Groq is raising $650 million for its new entities
Hot money rotation accelerates: from Bitcoin to gold, then to AI and storage semiconductors
Perplexity's open-source inference base pplx-garden bypasses Nvidia's network tax to achieve multi-card ultra-fast communication
NVIDIA released Gamma-World, a multi-agent world model supporting four-person collaboration and real-time 24 FPS
AI and leading tech stocks continue their strong momentum, while Gate expands its global asset trading coverage
SpaceX's self-developed C language AI training framework is nearly complete and will be used to train Grok v5
Foreign investors have bought Japanese stocks for the eighth consecutive week
US stock trading creates wealth on the chain: An address invested 4.3 million yuan in capital to enter the US stock market, with a 280% increase in two months.
"Photon Stock God": Nvidia's investment helps Nokia transform into AI infrastructure, with insiders actively increasing their holdings over the past month
Introducing AlphaGo search, the new MCTS video generation framework offers longer long video durations than Sora
Trending crypto
