Model Releases
NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning
NVIDIA’s open‑source Nemotron 3.5 Super VL is a 120 B parameter mixture‑of‑experts multimodal model that activates only 12 B parameters per token, enabling combined reasoning, instruction following, t
NVIDIA’s open‑source Nemotron 3.5 Super VL is a 120 B parameter mixture‑of‑experts multimodal model that activates only 12 B parameters per token, enabling combined reasoning, instruction following, tool use, and text, image, document, and video understanding within one framework. On a single GPU it achieves up to five times higher throughput for agentic workflows than comparable open models, while post‑training by organizations (Scale AI, Ramp, AI21) has yielded 40× lower inference cost, 10× lower latency, and 40 % higher accuracy. Developers can customize the model via NVIDIA NeMo using supervised fine‑tuning, low‑rank adaptation, or reinforcement learning, and deploy it efficiently with NVIDIA NIM, LiLiCorr speculative decoding, NVFP4 quantization, and other inference‑optimizations.
Related
- NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents
- NVIDIA Nemotron 3 Ultra Powers Faster, More Efficient Reasoning for Long-Running Agents
- .@NVIDIA Nemotron 3.5 Super VL is now live on Together AI. An open-weight multimodal model built for visual reasoning and always-on agents t…
- Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models
- Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for…
Source: NVIDIA Developer | 2026-10-15