Llama-3_1-Nemotron-Ultra-253B-V1 needs about 169.5 GB of memory at Q4_K_M (a 150.9 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 25 quants, from 54.9 GB (IQ1_S) to 506.8 GB (BF16).
Llama‑3.1‑Nemotron‑Ultra‑253B‑v1 – 253B parameter reasoning model
Llama‑3.1‑Nemotron‑Ultra‑253B‑v1 is a 253 billion parameter LLM optimized for reasoning, coding, and chat. It supports a context length of 128K tokens and is designed for inference on a single 8xH100 node. The model is released under the NVIDIA Open Model License.
Good at
- Reasoning tasks
- Coding assistance
- Tool‑calling
Limits
- Large memory requirement
- No vision support
- Requires NVIDIA Open Model License
Best for
- Enterprise RAG pipelines
- Complex code generation
- Chatbot integration
Who it’s for. Developers and data‑center operators needing high‑accuracy reasoning in a large‑context setting. The model is suitable for commercial deployments that can accommodate its memory footprint.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 54.9 GB (2 parts) | 61.9 GB | Too big | Get file → | |
| IQ1_M | 59.7 GB (2 parts) | 67.4 GB | Too big | Get file → | |
| IQ2_XXS | 68.1 GB (2 parts) | 76.7 GB | Too big | Get file → | |
| Q2_K | 93.4 GB (2 parts) | 105.1 GB | Too big | Get file → | |
| Q2_K_L | 93.9 GB (2 parts) | 105.7 GB | Too big | Get file → | |
| Q2_K_XL | 94.7 GB (2 parts) | 106.5 GB | Too big | Get file → | |
| Q3_K_S | 109.7 GB (3 parts) | 123.4 GB | Too big | Get file → | |
| Q3_K_M | 121.9 GB (3 parts) | 137.0 GB | Too big | Get file → | |
| Q3_K_XL | 123.2 GB (3 parts) | 138.5 GB | Too big | Get file → | |
| IQ4_XS | 135.5 GB (3 parts) | 152.2 GB | Too big | Get file → | |
| IQ4_NL | 143.2 GB (3 parts) | 160.9 GB | Too big | Get file → | |
| Q4_K_S | 144.4 GB (3 parts) | 162.2 GB | Too big | Get file → | |
| Q4_K_M | 150.9 GB (4 parts) | 169.5 GB | Too big | Get file → | |
| Q4_K_XL | 151.8 GB (4 parts) | 170.6 GB | Too big | Get file → | |
| Q4_1 | 158.8 GB (4 parts) | 178.4 GB | Too big | Get file → | |
| Q5_K_S | 174.5 GB (4 parts) | 195.9 GB | Too big | Get file → | |
| Q5_K_M | 178.5 GB (4 parts) | 200.5 GB | Too big | Get file → | |
| Q5_K_XL | 179.4 GB (4 parts) | 201.5 GB | Too big | Get file → | |
| IQ2_M | 192.7 GB (5 parts) | 216.3 GB | Too big | Get file → | |
| Q6_K | 207.9 GB (5 parts) | 233.3 GB | Too big | Get file → | |
| IQ3_XXS | 213.7 GB (5 parts) | 239.9 GB | Too big | Get file → | |
| Q6_K_XL | 215.0 GB (5 parts) | 241.3 GB | Too big | Get file → | |
| Q8_0 | 269.3 GB (6 parts) | 302.1 GB | Too big | Get file → | |
| Q8_K_XL | 281.0 GB (6 parts) | 315.2 GB | Too big | Get file → | |
| BF16 | 506.8 GB (11 parts) | 568.1 GB | Too big | Get file → |
For 8 GB VRAM use IQ1_S or IQ1_M. For 16 GB VRAM use Q2_K or Q2_K_L. For 32 GB VRAM use Q3_K_M or Q3_K_XL. Memory estimates are approximate and depend on context size.
What GPU do you need for Llama-3_1-Nemotron-Ultra-253B-V1?
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: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: no quant fits at an 8K context.
- 48 GB VRAM: no quant fits at an 8K context.
Which Llama-3_1-Nemotron-Ultra-253B-V1 quant should you choose?
Q4_K_M is the usual starting point for Llama-3_1-Nemotron-Ultra-253B-V1: it balances file size and quality well. Here it is a 150.9 GB file.
The Q8_0 file is 269.3 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_S at 54.9 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_1-Nemotron-Ultra-253B-V1
How much VRAM does Llama-3_1-Nemotron-Ultra-253B-V1 need?
About 169.5 GB at Q4_K_M with an 8K context window. That is the 150.9 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_1-Nemotron-Ultra-253B-V1 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 61.9 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Llama-3_1-Nemotron-Ultra-253B-V1 GGUF?
It depends on the quant: IQ1_S is 54.9 GB, IQ1_M is 59.7 GB, IQ2_XXS is 68.1 GB, Q2_K is 93.4 GB, Q2_K_L is 93.9 GB, Q2_K_XL is 94.7 GB and 19 more.
What context length does Llama-3_1-Nemotron-Ultra-253B-V1 support?
The GGUF header lists a trained context length of 131,072 tokens (128K).
Where can I download Llama-3_1-Nemotron-Ultra-253B-V1 GGUF?
The files are on Hugging Face in the unsloth/Llama-3_1-Nemotron-Ultra-253B-v1-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_1-Nemotron-Ultra-253B-V1 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 Llama‑3.1‑Nemotron‑Ultra‑253B‑v1 on 8 GB VRAM?
Only the IQ1_S or IQ1_M quantizations can fit on 8 GB VRAM, but they require 61.9 GB or 67.4 GB of memory at 8K context, so 8 GB VRAM is insufficient.
Does the model support vision inputs?
No, the model does not support vision or multimodal inputs.
What license governs the model?
The model is governed by the NVIDIA Open Model License and the Llama 3.1 Community License Agreement.