Nemotron 3.5 Bolt
NVIDIA · US · #13 in our ranking · Best on a 24GB GPU
Nemotron 3.5 Bolt is an open-weight mixture-of-experts model from NVIDIA (United States), released Oct 2026, with 74B parameters (9B active per token), a 256K-token context window, and the OpenMDW-1.1.
Bolt is a compressed version of Nemotron 3.5 Super VL. The BF16 weights are for fine-tuning, and the 24GB build is for local inference.
Nemotron 3.5 Bolt specs
- Provider
- NVIDIA
- Origin
- US
- Architecture
- Mixture-of-experts
- Parameters
- 74B total · 9B active
- Context
- 256K tokens | 197K words
- License
- OpenMDW-1.1 Commercial use
- Input
- Text, Image, Video
- Released
- Oct 2026
- Weights
- Hugging Face
Sourced from the model card on Hugging Face. Last checked 2026-10-16.
Highlights
- 74B total, 9B active
- 24GB build for local GPUs
- Permissive OpenMDW license
Where it fits for payment teams
Worth testing for high-volume, low-risk steps on a single workstation GPU.
Harnesses in Runtime
Run Nemotron 3.5 Bolt through any of these harnesses on Runtime.
Provider-reported benchmarks
| Terminal-Bench 2.1 | 53.9 |
|---|---|
| SciCode | 44.5 |
| HLE (no tools) | 28.7 |
Benchmark scores are reported by each model provider on its Hugging Face model card. They are not independently verified by Runtime.
How to run Nemotron 3.5 Bolt
Download the weights
Pull nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16 from Hugging Face. Check the OpenMDW-1.1.
Serve it in your cloud
vLLM or SGLang on your GPUs, or a managed host.
Put it to work in Runtime
Point an agent or skill at it, then compare with evals.
Nemotron 3.5 Bolt FAQ
What is the context window of Nemotron 3.5 Bolt?
+
Nemotron 3.5 Bolt supports a 256K-token context window (262,144 tokens), roughly 197K words of English text.
How many parameters does Nemotron 3.5 Bolt have?
+
Nemotron 3.5 Bolt is a mixture-of-experts model with 74B total parameters, of which 9B are active per token.
Is Nemotron 3.5 Bolt free for commercial use?
+
Yes. Nemotron 3.5 Bolt is released under the OpenMDW-1.1 license, which is permissive and allows commercial use and fine-tuning.
What can Nemotron 3.5 Bolt take as input?
+
Nemotron 3.5 Bolt accepts text, image, video input and generates text.
Can I run Nemotron 3.5 Bolt in my own cloud with Runtime?
+
Yes. Download the weights from Hugging Face (nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16) and serve them with vLLM or SGLang in your cloud, or use a managed provider that hosts the model. Runtime agents can then use Nemotron 3.5 Bolt through a harness like OpenCode, with your data staying in your environment.