What are the hardware requirements for Llama 3 1 Nemotron Ultra 253B CPT V1?

Llama 3 1 Nemotron Ultra 253B CPT V1 requires approximately 167.6 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Llama 3 1 Nemotron Ultra 253B CPT V1 and how large is it?

Llama 3 1 Nemotron Ultra 253B CPT V1 is developed by nvidia, a model with 253b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Llama 3 1 Nemotron Ultra 253B CPT V1?

Llama 3 1 Nemotron Ultra 253B CPT V1 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Llama 3 1 Nemotron Ultra 253B CPT V1 by nvidia — VRAM 167.6GB, 128K context

Name: Llama 3 1 Nemotron Ultra 253B CPT V1
Author: nvidia

Llama 3 1 Nemotron Ultra 253B CPT V1 is an open-source language model by nvidia. Features: 253b LLM, VRAM: 167.6GB, Context: 128K, License: other, LLM Explorer Score: 0.18.

Arxiv:2411.19146 Arxiv:2503.18908 Conversational Custom code En Llama-3 Nemotron-nas Nvidia Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1

Llama 3 1 Nemotron Ultra 253B CPT V1 Parameters and Internals

LLM Name	Llama 3 1 Nemotron Ultra 253B CPT V1
Repository 🤗	https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1
Model Size	253b
Required VRAM	167.6 GB
Updated	2026-08-08
Maintainer	nvidia
Model Type	nemotron-nas
Model Files	4.9 GB: 1-of-96 5.0 GB: 2-of-96 4.8 GB: 3-of-96 4.9 GB: 4-of-96 4.7 GB: 5-of-96 3.2 GB: 6-of-96 4.9 GB: 7-of-96 4.7 GB: 8-of-96 4.4 GB: 9-of-96 3.5 GB: 10-of-96 4.6 GB: 11-of-96 4.6 GB: 12-of-96 3.5 GB: 13-of-96 4.6 GB: 14-of-96 4.6 GB: 15-of-96 3.5 GB: 16-of-96 4.6 GB: 17-of-96 4.6 GB: 18-of-96 3.5 GB: 19-of-96 4.6 GB: 20-of-96 4.6 GB: 21-of-96 4.5 GB: 22-of-96 4.7 GB: 23-of-96 4.4 GB: 24-of-96 3.9 GB: 25-of-96 4.6 GB: 26-of-96 3.5 GB: 27-of-96 3.5 GB: 28-of-96 4.6 GB: 29-of-96 3.5 GB: 30-of-96 4.6 GB: 31-of-96 4.6 GB: 32-of-96 3.5 GB: 33-of-96 4.6 GB: 34-of-96 4.6 GB: 35-of-96 3.5 GB: 36-of-96 4.6 GB: 37-of-96 4.6 GB: 38-of-96 3.5 GB: 39-of-96
Supported Languages	en
Model Architecture	DeciLMForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.48.3
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Llama 3 1 Nemotron Ultra 253B CPT V1 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Llama 3 1 Nemotron Ultra 253B CPT V1

Best Alternatives	Context / RAM	Downloads	Likes
...ama 3 1 Nemotron Ultra 253B V1	128K / 405.3 GB	3094	352
...3 1 Nemotron Ultra 253B V1 FP8	128K / 198.8 GB	732	12
... 1 Nemotron Ultra 253B V1 GGUF	128K / GB	1604	10

Note: green Score (e.g. "73.2") means that the model is better than nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1.