NVIDIA Launches Nemotron 3.5 Series: From Data Center to Desktop Efficiency
Summary:NVIDIA has unveiled the Nemotron 3.5 series, featuring the 120-billion-parameter Super VL model and two new edge-optimized models: Bolt 74B-A9B 24GB and Bolt E4B 4GB. The Super VL model excels in multimodal agentic tasks, delivering high accuracy and efficiency, while the Bolt series is optimized for local agentic workloads, balancing intelligence and speed. The Nemotron 3.5 series supports deployment across enterprise servers to edge devices and allows users to fine-tune and customize models to
Nemotron 3.5 Super VL: A Leap in Multimodal Agentic Performance
Nemotron 3.5 Super VL, developed in collaboration with the Nemotron Coalition, is an open-source model with 120 billion parameters that significantly enhances accuracy and efficiency in multimodal agentic tasks. Key highlights include:
- High Accuracy and Efficiency: Super VL achieves accuracy comparable to NVIDIA's largest reasoning models but with only one-fifth of the parameters, making it highly efficient for complex tasks.
- Task Completion Speed: As the fastest open agentic model in its class, Super VL is 5x faster than the next best, significantly improving the throughput of enterprise agents across workflows.
- Open and Customizable: Users can fine-tune the model on their own data and deploy it on their infrastructure, ensuring data privacy and security.
Super VL excels on NVIDIA's AI analysis index leaderboard, outperforming other models in throughput. It also supports multimodal understanding of complex documents like webpages and PDFs, extracting insights and making decisions directly from visual and textual information.
Nemotron 3.5 Bolt: Optimized for Edge Intelligence
The Bolt series is designed for edge computing and includes:
- Bolt 74B-A9B 24GB: Optimized for local agentic workloads, it uses 2-bit quantization to fit in 24GB of memory and is suitable for everyday intelligent tasks such as web research, writing, and code debugging.
- Bolt E4B 4GB: Optimized for fast conversational agents, it combines natural language conversation with multi-turn agentic capabilities, powering local voice assistants, interactive characters, and embedded interfaces.
Bolt 74B-A9B 24GB is the top open model in its class on PinchBench and BrowseComp benchmarks.
Deployment and Customization
Nemotron 3.5 models support customization through NVIDIA NeMo, Python pipelines, and open datasets. NVIDIA NIM packages the model as a production-ready microservice with performance-optimized, validated configurations and enterprise-grade support, refreshed every two weeks with security patches and performance improvements.
Hardware Support
Nemotron 3.5 Super VL runs across NVIDIA's full hardware stack, including DGX Spark, DGX Station, RTX PRO workstations, GeForce RTX PCs, and NVIDIA Jetson edge devices. Bolt 74B-A9B 24GB supports multiple hardware platforms, including the upcoming RTX Spark laptops.
Conclusion
The release of the Nemotron 3.5 series marks NVIDIA's continued innovation in AI models, providing efficient and flexible solutions for a wide range of applications from data centers to edge devices. Its openness and customizability make it an ideal choice for enterprises building intelligent agents.
— END —Source: NVIDIA AI News (2026-10-15)
Tags: #NVIDIA #Nemotron #Multimodal Models #Edge Computing #AI Models
Editorial & Fact-Checking Note: This article is compiled from primary research, official release documentation, and source papers by the ZICQ Newsroom pipeline with automated entity verification and human editorial review. If you notice any technical inaccuracy, please submit a correction via our corrections policy or email our editorial desk directly.
Community Comments