Picture published by the creator on Hugging Face.
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16: Ask about documents and video
Added · Updated
Made by NVIDIA
View profile on Hugging Face (opens in a new tab)Ask questions about document images, pictures and video and get text answers; it needs NVIDIA GPU servers, so it suits developers.
Model details
Original model nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16
Input Image or text Output Text
Ways to use it Run locally
What it does
This is a vision language model: it reads text, images and video and answers in text. NVIDIA lists reasoning, coding, tool use, document understanding and video understanding as its strengths. It has 120B total parameters, of which 12B are active, and supports context lengths up to 1M tokens. Reasoning can be switched on or off through the chat template.
How to use it
Input → action → output: start a model server from the Quick Start, then send a question together with a document image, picture or video through the documented client examples, and receive a written text answer. Before sending anything, wait for the server to finish loading and check that its endpoint responds.
Common use cases
- Send a picture of a document page with a question about it and read a text answer.
- Send a video and ask for a written explanation of what it shows.
Can I use this?
Requirements Use Docker with NVIDIA GPU support and the vLLM server image named on the model page. NVIDIA lists 4× H100 or 2× GB200 for deployment; the Quick Start example serves the model on 4× H100. If the model asks for authentication, set a Hugging Face token (HF_TOKEN) before launching the container. The license is OpenMDW 1.1; the source says the model is ready for commercial or non-commercial use.
Caveat Hardware beyond the listed NVIDIA servers is not documented. The medium reasoning setting uses fewer tokens than the on setting but may reduce accuracy. NVIDIA points to a separate NVFP4 model for optimized deployment; these BF16 files are the full-precision weights.
Worth knowing The medium reasoning setting generates fewer tokens than the on setting but may reduce accuracy. Licensed under OpenMDW 1.1.
Why it's here
Not trending this week
Last seen on the Hugging Face trending list on 2 November 2026. It stays on BuildTube so you can still find it.
- 117 people have liked it on Hugging Face.
- It was downloaded 71,232 times in the last 30 days.
Numbers checked on 8 November 2026; not refreshed since.
Behind it
View model pageBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16?
- This is a vision language model: it reads text, images and video and answers in text.
- What can I use NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 for?
- Send a picture of a document page with a question about it and read a text answer; Send a video and ask for a written explanation of what it shows
- Can I run NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 on my computer?
- At about 120 billion parameters in total (12 billion active), it needs a multi-GPU server, not a desktop or laptop. Use Docker with NVIDIA GPU support and the vLLM server image named on the model page. NVIDIA lists 4× H100 or 2× GB200 for deployment; the Quick Start example serves the model on 4× H100. Hardware beyond the listed NVIDIA servers is not documented.
- What do I need to use NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16?
- Docker with NVIDIA GPU support, the documented vLLM server image, 4× H100 or 2× GB200 for deployment, and acceptance of the OpenMDW 1.1 license.
- What license does NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 use?
- Licensed under OpenMDW 1.1. The license is OpenMDW 1.1; the source says the model is ready for commercial or non-commercial use.
More from this week
- NVKE Run managed Kubernetes clusters on your own NVIDIA GPU machines
Set up a service that creates and manages separate Kubernetes clusters on your own NVIDIA GPU servers.
- P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
- Idol Turn one picture of an object into a 3D Gaussian splat
Idol turns one photo, render or generated image of an object into a 3D Gaussian splat file.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/hf-nvidia-nvidia-nemotron-3-5-super-vl-120b-a12b-bf16/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/hf-nvidia-nvidia-nemotron-3-5-super-vl-120b-a12b-bf16/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 on BuildTube"></iframe>