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

Llama 3 1 Nemotron Ultra 253B V1 requires approximately 405.3 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 V1 and how large is it?

Llama 3 1 Nemotron Ultra 253B 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.

How does Llama 3 1 Nemotron Ultra 253B V1 perform on standard benchmarks?

Llama 3 1 Nemotron Ultra 253B V1 has the following published scores: LMSYS ELO 1348. Compare against reference models on this page or on the LLM Explorer leaderboards.

Llama 3 1 Nemotron Ultra 253B V1 by nvidia — VRAM 405.3GB, 128K context

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

Llama 3 1 Nemotron Ultra 253B V1 is an open-source language model by nvidia. Features: 253b LLM, VRAM: 405.3GB, Context: 128K, License: other, LLM Explorer Score: 0.36, ELO: 1348.

Arxiv:2411.19146 Arxiv:2502.00203 Arxiv:2503.18908 Arxiv:2505.00949 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-v1

Llama 3 1 Nemotron Ultra 253B V1 Benchmarks

LMSys ELO 1348 vs 1272 (so35) 6%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Llama 3 1 Nemotron Ultra 253B V1 Parameters and Internals

LLM Name	Llama 3 1 Nemotron Ultra 253B V1
Repository 🤗	https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-v1
Model Size	253b
Required VRAM	405.3 GB
Updated	2026-08-05
Maintainer	nvidia
Model Type	nemotron-nas
Model Files	9.5 GB: 1-of-49 9.9 GB: 2-of-49 9.4 GB: 3-of-49 10.0 GB: 4-of-49 9.5 GB: 5-of-49 8.8 GB: 6-of-49 7.5 GB: 7-of-49 8.1 GB: 8-of-49 8.7 GB: 9-of-49 9.8 GB: 10-of-49 8.3 GB: 11-of-49 9.6 GB: 12-of-49 7.5 GB: 13-of-49 9.9 GB: 14-of-49 8.1 GB: 15-of-49 7.5 GB: 16-of-49 9.9 GB: 17-of-49 8.1 GB: 18-of-49 7.5 GB: 19-of-49 9.9 GB: 20-of-49 8.1 GB: 21-of-49 7.5 GB: 22-of-49 9.9 GB: 23-of-49 8.1 GB: 24-of-49 9.9 GB: 25-of-49 8.4 GB: 26-of-49 9.5 GB: 27-of-49 9.1 GB: 28-of-49 9.1 GB: 29-of-49 9.3 GB: 30-of-49 9.9 GB: 31-of-49 8.7 GB: 32-of-49 9.9 GB: 33-of-49 6.7 GB: 34-of-49 26.2 GB: 35-of-49 13.1 GB: 36-of-49 27.9 GB: 37-of-49 14.0 GB: 38-of-49 22.5 GB: 39-of-49
Supported Languages	en
Model Architecture	DeciLMForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.45.1
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Llama 3 1 Nemotron Ultra 253B 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

Quantized Models of the Llama 3 1 Nemotron Ultra 253B V1

Model	Likes	Downloads	VRAM
... 1 Nemotron Ultra 253B V1 GGUF	10	1604	0 GB

Best Alternatives to Llama 3 1 Nemotron Ultra 253B V1

Best Alternatives	Context / RAM	Downloads	Likes
...3 1 Nemotron Ultra 253B V1 FP8	128K / 198.8 GB	732	12
...3 1 Nemotron Ultra 253B CPT V1	128K / 167.6 GB	101	6
... 1 Nemotron Ultra 253B V1 GGUF	128K / GB	1604	10