Llama 3_3 Nemotron Super 49B GenRM Multilingual needs about 34.3 GB of memory at Q4_K_M (a 30.2 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 26 quants, from 12.0 GB (IQ1_M) to 99.7 GB (BF16).
Llama‑3.3 Nemotron Super 49B GenRM Multilingual – a reward model for ranking responses
Llama‑3.3‑Nemotron‑Super‑49B‑GenRM‑Multilingual is a generative reward model built on Llama‑3.3‑Nemotron‑Super‑49B‑v1. It evaluates the quality of LLM responses, producing reasoning traces and an integer score. The model is available for commercial and non‑commercial use.
Good at
- Evaluating single response quality
- Ranking two responses in a multilingual context
- Generating reasoning traces before scoring
Limits
- No vision support
- Requires large VRAM for full precision
- License governed by NVIDIA Open Model License
Best for
- LLM-as‑judge applications
- Quality assessment in multilingual chat
- Benchmarking reward models
Who it’s for. Developers building LLM‑as‑judge systems or evaluating response quality. Researchers needing a multilingual reward model for benchmarking. Teams that require a model to rank or score LLM outputs.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_M | 12.0 GB | 14.0 GB | Too big | Get file → | |
| IQ2_XXS | 13.7 GB | 15.8 GB | Too big | Get file → | |
| IQ2_XS | 15.1 GB | 17.4 GB | Too big | Get file → | |
| IQ2_S | 15.8 GB | 18.3 GB | Too big | Get file → | |
| IQ2_M | 17.2 GB | 19.7 GB | Too big | Get file → | |
| Q2_K | 18.7 GB | 21.5 GB | Too big | Get file → | |
| IQ3_XXS | 19.5 GB | 22.4 GB | Too big | Get file → | |
| Q2_K_L | 19.8 GB | 22.6 GB | Too big | Get file → | |
| IQ3_XS | 20.9 GB | 23.9 GB | Too big | Get file → | |
| Q3_K_S | 22.0 GB | 25.1 GB | Too big | Get file → | |
| IQ3_M | 22.7 GB | 25.9 GB | Too big | Get file → | |
| Q3_K_M | 24.3 GB | 27.7 GB | Too big | Get file → | |
| Q3_K_L | 26.3 GB | 29.9 GB | Too big | Get file → | |
| IQ4_XS | 26.9 GB | 30.6 GB | Too big | Get file → | |
| Q3_K_XL | 27.2 GB | 31.0 GB | Too big | Get file → | |
| IQ4_NL | 28.4 GB | 32.3 GB | Too big | Get file → | |
| Q4_0 | 28.5 GB | 32.4 GB | Too big | Get file → | |
| Q4_K_S | 28.6 GB | 32.6 GB | Too big | Get file → | |
| Q4_K_M | 30.2 GB | 34.3 GB | Too big | Get file → | |
| Q4_K_L | 31.0 GB | 35.2 GB | Too big | Get file → | |
| Q4_1 | 31.4 GB | 35.6 GB | Too big | Get file → | |
| Q5_K_S | 34.4 GB | 39.1 GB | Too big | Get file → | |
| Q5_K_M | 35.4 GB | 40.1 GB | Too big | Get file → | |
| Q6_K | 40.9 GB | 46.3 GB | Too big | Get file → | |
| Q8_0 | 53.0 GB (2 parts) | 59.9 GB | Too big | Get file → | |
| BF16 | 99.7 GB (3 parts) | 112.2 GB | Too big | Get file → |
For 8 GB VRAM use IQ1_M (file size 12.02 GB, memory 14 GB). With 16 GB VRAM IQ2_S (15.85 GB, 18.3 GB) is suitable. For 32 GB VRAM or more, Q4_K_M (30.22 GB, 34.3 GB) or Q4_K_L (31 GB, 35.2 GB) provide a balance of size and quality.
What GPU do you need for Llama 3_3 Nemotron Super 49B GenRM Multilingual?
These figures add the file size, an fp16 KV cache at an 8K context window 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 IQ2_XXS (13.7 GB file, about 15.8 GB in use).
- 24 GB VRAM: the best quant that fits is IQ3_XS (20.9 GB file, about 23.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q3_K_XL (27.2 GB file, about 31.0 GB in use).
- 48 GB VRAM: the best quant that fits is Q6_K (40.9 GB file, about 46.3 GB in use).
Which Llama 3_3 Nemotron Super 49B GenRM Multilingual quant should you choose?
Q4_K_M is the usual starting point for Llama 3_3 Nemotron Super 49B GenRM Multilingual: it balances file size and quality well. Here it is a 30.2 GB file.
The Q8_0 file is 53.0 GB, about 1.8 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 IQ1_M at 12.0 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about Llama 3_3 Nemotron Super 49B GenRM Multilingual
How much VRAM does Llama 3_3 Nemotron Super 49B GenRM Multilingual need?
About 34.3 GB at Q4_K_M with an 8K context window. That is the 30.2 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 Llama 3_3 Nemotron Super 49B GenRM Multilingual on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 14.0 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Llama 3_3 Nemotron Super 49B GenRM Multilingual GGUF?
It depends on the quant: IQ1_M is 12.0 GB, IQ2_XXS is 13.7 GB, IQ2_XS is 15.1 GB, IQ2_S is 15.8 GB, IQ2_M is 17.2 GB, Q2_K is 18.7 GB and 20 more.
What context length does Llama 3_3 Nemotron Super 49B GenRM Multilingual support?
The GGUF header lists a trained context length of 131,072 tokens (128K).
Where can I download Llama 3_3 Nemotron Super 49B GenRM Multilingual GGUF?
The files are on Hugging Face in the bartowski/nvidia_Llama-3_3-Nemotron-Super-49B-GenRM-Multilingual-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 Llama 3_3 Nemotron Super 49B GenRM Multilingual 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 this model on 8 GB VRAM?
Yes, the IQ1_M quant requires 14 GB of VRAM at 8 k context, which is close to 8 GB but may need swapping. Smaller quants are not available.
Does this model support vision inputs?
No, the model does not include vision capabilities.
What license applies to this model?
The model is governed by the NVIDIA Open Model License and the Llama 3.3 Community License Agreement.