NVIDIA releases Nemotron 3.5 Super VL with day 0 vLLM support
Original titleAnnouncing Day 0 Support for NVIDIA Nemotron 3.5 Super VL on vLLM
AISummary
NVIDIA's Nemotron Team announces Nemotron 3.5 Super VL, a 120B-parameter mixture-of-experts model with 12B active parameters that handles text, images, documents, and video as input.
The model is available on Hugging Face in BF16 and NVFP4 and can be served with vLLM v0.31.0 on listed NVIDIA hardware. The post says it runs up to 5x faster than comparable open models on agentic tasks, according to its charts.
AIWhy it matters
The post gives the model's architecture, deployment targets, and vLLM serving commands, which helps engineers judge whether it fits their hardware and agent workloads.
Source: vLLM Blog · vllm.aiPublished · added here