Nemotron 3.5 Super VL
Open 74B vision-language model that runs locally on a 24GB GPU
NVIDIA's open-weight vision-language model (74B MoE, 9B active) that runs on a single 24GB GPU.
Model card on Hugging Face
1 / 2Overview
What is Nemotron 3.5 Super VL?
Open 74B vision-language model that runs locally on a 24GB GPU
Nemotron 3.5 Super VL is NVIDIA's open-weight vision-language model: 74B parameters in a mixture-of-experts architecture with only 9B active per token, a 256K context window, and 2-bit quantization that lets it run locally on a single 24GB GPU.
It brings big-model vision-language understanding — images and text together, long documents, multi-image reasoning — to local RTX workstations and DGX Spark, without cloud APIs.
Released under the OpenMDW-1.1 license for research and commercial use.
Platforms and languages
Nemotron 3.5 Super VL Availability
Platforms
Languages
Capabilities
Nemotron 3.5 Super VL Key Features
74B MoE, 9B active
74 billion total parameters with only 9B active per token — big-model capability at a fraction of the compute.
Runs on a 24GB GPU
2-bit quantization squeezes the model onto a single 24GB card: local RTX workstations and DGX Spark.
Vision + language
A true vision-language model: understands images and text together.
256K context
Long-context window for documents, code and multi-image reasoning.
Open weights
Released under the OpenMDW-1.1 license for research and commercial use.
Best for
Who uses Nemotron 3.5 Super VL?
AI developers
Run a capable open vision-language model locally on consumer GPUs
Plans and access
Nemotron 3.5 Super VL Pricing
Free
Common questions
Nemotron 3.5 Super VL FAQs
What hardware do I need?
Thanks to 2-bit quantization, it runs on a single 24GB GPU such as an RTX card or NVIDIA DGX Spark.
Is it really open?
Yes — open weights under the OpenMDW-1.1 license.
How big is the model?
74B total parameters in a mixture-of-experts design, with 9B active per token, and a 256K context window.
Reviews
See what the community thinks and share your experience.
—
Based on 0 ratings
Rating distribution
Community reviews
No reviews
No written reviews yet.
