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
Loeb refutes the bubble theory: investing in AI is definitely not about flushing toilets with money; we have only just touched the surface of the hype
Zhiyuan Robotics: The 2-billion-parameter model GE 2.0 topped the international review rankings
"New Stock God" Serenity Predicts AI Hardware Investment Opportunities: Foxconn CPO Switch Ships in Q3, Xunxin Responsible for Its Advanced Optics Business
Aethir has officially launched the delivery of computing power for Axe Compute's $260 million B300 cluster, with the first $43 million payment already in place
The top ten US companies are now all occupied by technology and AI-related firms
"New stock god" Serenity is bullish on the AI photonics industry chain, naming AAOI, SIVE, Foci, and Shunsin
NVIDIA spends over $6.5 billion on photonic technology, increasing investment in AI infrastructure "data transmission bottlenecks"
The AI supercycle has created a computing power gap, with Render, Bittensor, and others as major beneficiaries
US stocks in semiconductors, memory chips, servers, and application software all surged, with multiple stocks hitting new all-time highs
"BTC OG Insider Whale" Agent: AI Drains Crypto Market Liquidity, The Next Bull Market Still Needs to Wait for Cyclical Return
