Nemotron 3 Super 120B A12B needs about 99.3 GB of memory at Q4_K_M (a 87.0 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 28 quants, from 0.3 GB to 128.5 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 47.2 GB.
Nemotron 3 Super 120B A12B – large LLM with 12B active parameters and 120B total
Nemotron 3 Super 120B A12B is a large language model trained by NVIDIA. It uses a hybrid nemotron_h_moe architecture with 12B active parameters and a 1,048,576 token context window. The model is available in multiple GGUF quantizations for different memory budgets.
Good at
- High‑volume agentic tasks
- Reasoning with multi‑token prediction
- Large context handling
Limits
- Requires 8‑k context for memory estimates
- No vision support
- License is NVIDIA Nemotron Open Model License
Best for
- IT ticket automation
- Collaborative agent workflows
- High‑volume text generation
Who it’s for. Developers building AI agents that need strong reasoning and conversational abilities. The model is ready for commercial use and supports multiple languages such as English, French, German, Italian, Japanese, Spanish, and Chinese.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.3 GB | 12.6 GB | Too big | Get file → | |
| IQ1_S | 46.4 GB | 58.7 GB | Too big | Get file → | |
| IQ1_M | 47.8 GB | 60.1 GB | Too big | Get file → | |
| IQ2_XXS | 50.2 GB (2 parts) | 62.5 GB | Too big | Get file → | |
| IQ2_XS | 52.0 GB (2 parts) | 64.4 GB | Too big | Get file → | |
| IQ2_S | 52.1 GB (2 parts) | 64.4 GB | Too big | Get file → | |
| IQ2_M | 54.0 GB (2 parts) | 66.3 GB | Too big | Get file → | |
| Q2_K | 54.8 GB (2 parts) | 67.1 GB | Too big | Get file → | |
| Q2_K_L | 55.3 GB (2 parts) | 67.7 GB | Too big | Get file → | |
| Q3_K_S | 61.1 GB (2 parts) | 73.4 GB | Too big | Get file → | |
| IQ3_XXS | 61.6 GB (2 parts) | 73.9 GB | Too big | Get file → | |
| IQ3_XS | 64.5 GB (2 parts) | 76.9 GB | Too big | Get file → | |
| Q3_K_M | 64.7 GB (2 parts) | 77.0 GB | Too big | Get file → | |
| IQ3_M | 66.3 GB (2 parts) | 78.6 GB | Too big | Get file → | |
| Q3_K_L | 66.4 GB (2 parts) | 78.7 GB | Too big | Get file → | |
| Q3_K_XL | 66.9 GB (2 parts) | 79.2 GB | Too big | Get file → | |
| IQ4_XS | 67.2 GB (2 parts) | 79.5 GB | Too big | Get file → | |
| IQ4_NL | 69.1 GB (2 parts) | 81.4 GB | Too big | Get file → | |
| Q4_0 | 71.0 GB (2 parts) | 83.3 GB | Too big | Get file → | |
| Q4_1 | 76.6 GB (2 parts) | 88.9 GB | Too big | Get file → | |
| Q4_K_S | 78.0 GB (2 parts) | 90.4 GB | Too big | Get file → | |
| Q4_K_M | 87.0 GB (3 parts) | 99.3 GB | Too big | Get file → | |
| Q4_K_L | 87.4 GB (3 parts) | 99.7 GB | Too big | Get file → | |
| Q5_K_S | 87.6 GB (3 parts) | 99.9 GB | Too big | Get file → | |
| Q5_K_M | 96.6 GB (3 parts) | 108.9 GB | Too big | Get file → | |
| Q5_K_L | 96.9 GB (3 parts) | 109.2 GB | Too big | Get file → | |
| Q6_K | 113.6 GB (3 parts) | 125.9 GB | Too big | Get file → | |
| Q8_0 | 128.5 GB (4 parts) | 140.8 GB | Too big | Get file → |
For 8‑GB VRAM, use IQ1_S (file size 46.38 GB, memory 58.7 GB at 8k context). For 16‑GB VRAM, Q4_0 (70.99 GB, 83.3 GB). For 24‑GB VRAM, Q4_K_M (86.98 GB, 99.3 GB). Estimates are approximate.
What GPU do you need for Nemotron 3 Super 120B A12B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 12.6 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 12.6 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 12.6 GB in use).
- 48 GB VRAM: the best quant that fits is the main file (0.3 GB file, about 12.6 GB in use).
Which Nemotron 3 Super 120B A12B quant should you choose?
Q4_K_M is the usual starting point for Nemotron 3 Super 120B A12B: it balances file size and quality well. Here it is a 87.0 GB file.
The Q8_0 file is 128.5 GB, about 1.5 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is the first one in the table at 0.3 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about Nemotron 3 Super 120B A12B
How much VRAM does Nemotron 3 Super 120B A12B need?
About 99.3 GB at Q4_K_M with an 8K context window. That is the 87.0 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Nemotron 3 Super 120B A12B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 12.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Nemotron 3 Super 120B A12B GGUF?
It depends on the quant: file is 0.3 GB, IQ1_S is 46.4 GB, IQ1_M is 47.8 GB, IQ2_XXS is 50.2 GB, IQ2_XS is 52.0 GB, IQ2_S is 52.1 GB and 22 more.
What context length does Nemotron 3 Super 120B A12B support?
The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 11.8 GB at 8K and 47.2 GB at 32K.
Where can I download Nemotron 3 Super 120B A12B GGUF?
The files are on Hugging Face in the bartowski/nvidia_Nemotron-3-Super-120B-A12B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Nemotron 3 Super 120B A12B use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Nemotron 3 Super 120B A12B on 8 GB VRAM?
No, the smallest quant requires 46.38 GB file size and 58.7 GB memory at 8k context. 8 GB VRAM is insufficient.
Does the model support vision tasks?
No, the model has no vision support.
What license applies to this model?
The NVIDIA Nemotron Open Model License governs use of the model.