Runs on a private AI box you own
One compact GB10-class device — 128 GB unified memory — runs it on-premise, sold and provisioned by us. No cloud API keys, no data leaving the building.
Run Llama-Nemotron Super 49B on-premises
○ Memory-fit arithmetic — checkable math, not a deployment claim
Llama-Nemotron Super 49B is NVIDIA's reasoning- and tool-calling-tuned model, NAS-compressed specifically to run well on a single accelerator. On an NVIDIA GB10-class box — exactly the hardware family it targets — it fits with plenty of headroom.
Here is the arithmetic on a 128 GB unit.
Memory fit on a 128 GB unit
| Parameters | 49B (dense, NAS-compressed) |
|---|---|
| 4-bit weights | ~28 GB |
| Device memory | 128 GB unified (one NVIDIA GB10-class unit) |
| Left for context + other models | roughly 90 GB |
| Units required | 1 |
The model was compressed to fit a single accelerator, so a 128 GB unit runs it with large context headroom to spare.
What it serves well
- Reasoning, RAG, and tool-calling workloads on NVIDIA hardware.
- Boxes that want NVIDIA-native tuning and ecosystem alignment.
- Leaving memory free to co-host embedding and coder models.
Honest limits
- Arithmetic, not our measured deployment.
- It ships under the NVIDIA Open Model License (with Llama 3.3 community terms); we review licensing fit in the assessment.
- The larger Nemotron Ultra (~253B) does not comfortably fit one 128 GB unit.
Frequently asked questions
- Does Super 49B fit one unit?
- Yes, comfortably — ~28 GB at 4-bit against 128 GB leaves roughly 90 GB for context and other models.
- Why is it a good NVIDIA-box fit?
- It was NAS-compressed by NVIDIA to run on a single accelerator, and our devices are NVIDIA GB10-class systems — the family it is tuned for.
- License?
- NVIDIA Open Model License with Llama 3.3 community terms; we confirm fit for your use in the assessment.