Nemotron 3.5 Bolt
Nemotron 3.5 Bolt is an open-weight vision-language model from NVIDIA, released on October 15, 2026.
- Developer
- NVIDIA
- Released
- October 15, 2026
- Total parameters
- 74B
- Active parameters per token
- 9B
- Architecture
- MoE
- Context window
- 256K
- Input
- Text, image, video
- License
- OpenMDW License Agreement v1.1
- Commercial use
- Yes
- Official weights, GB
- 149.8 GB
- Hugging Face repository
- huggingface.co
- Checked on
Nemotron 3.5 Bolt is a mixture-of-experts model with 74B total parameters, of which 9B are active for each token. NVIDIA lists a context window of up to 256K tokens. The model takes text, image and video input and produces text. The weights are released under the OpenMDW License Agreement v1.1, which allows commercial use.
The official BF16 weights on Hugging Face take about 149.8 GB. NVIDIA describes this release as intended mainly for customization and post-training. Running the model needs at least that much memory across GPUs and system RAM, plus room for the context cache; NVIDIA’s smaller variants need less.
- Total parameters, billions
- 74B
- Active parameters, billions
- 9B
- Context window, tokens
- 256K
- License type
- Custom license
- License conditions
- OpenMDW 1.1: redistributions must include the agreement and origin notices.
- Weights precision
- BF16
- Official quantized or alternative versions
- nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB, nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-LiLiCorr
- License text
- raw.githubusercontent.com
- Sources
Page URL Hugging Face model card https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/main/README.md Hugging Face weights index https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/main/model.safetensors.index.json
Questions
What license does Nemotron 3.5 Bolt use?
How much memory does Nemotron 3.5 Bolt need?
How many parameters does Nemotron 3.5 Bolt have?
What context window does Nemotron 3.5 Bolt have?
Sources
Every figure comes from the developer’s model card, license text and official announcements, and from the model’s Hugging Face repository metadata. Each model page lists its sources and the date they were checked; the register is rechecked every week.