What are the hardware requirements for Minitron 4B Base?

Minitron 4B Base requires approximately 8.4 GB of VRAM and supports a context window of 4K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Minitron 4B Base and how large is it?

Minitron 4B Base is developed by nvidia, a model with 4b 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 Minitron 4B Base?

Minitron 4B Base is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Minitron 4B Base by nvidia — VRAM 8.4GB, 4K context

Name: Minitron 4B Base
Author: nvidia

Minitron 4B Base is an open-source language model by nvidia. Features: 4b LLM, VRAM: 8.4GB, Context: 4K, License: other, LLM Explorer Score: 0.19.

Arxiv:2009.03300 Arxiv:2407.14679 En Endpoints compatible Llama-3 Nemo Nemotron Nvidia Pytorch Region:us

Model Card on HF 🤗: https://huggingface.co/nvidia/Minitron-4B-Base

Minitron 4B Base Parameters and Internals

Model Type

Transformer Decoder, auto-regressive language model

Use Cases

Areas:

research, development

Limitations:

amplifies biases, can generate toxic responses, may produce inaccurate or undesirable text

Considerations:

Work with internal model team to ensure model meets industry requirements and addresses unforeseen misuse.

Supported Languages

en (basic proficiency), multilingual (basic proficiency)

Training Details

Data Sources:

webpages, dialogue, articles, other written materials

Data Volume:

94 billion tokens

Methodology:

pruning and knowledge distillation

Model Architecture:

Transformer Decoder

Input Output

Input Format:

String

Accepted Modalities:

Text

Output Format:

String

Performance Tips:

Load the model using TensorRT-LLM on supported hardware with CUDA and Torch settings for optimal performance.

LLM Name	Minitron 4B Base
Repository 🤗	https://huggingface.co/nvidia/Minitron-4B-Base
Model Size	4b
Required VRAM	8.4 GB
Updated	2026-07-29
Maintainer	nvidia
Model Type	nemotron
Model Files	8.4 GB
Supported Languages	en
Model Architecture	NemotronForCausalLM
License	other
Context Length	4096
Model Max Length	4096
Transformers Version	4.32.0.dev0
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	256000
Torch Data Type	bfloat16

Rank the Minitron 4B Base 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 Minitron 4B Base

Best Alternatives	Context / RAM	Downloads	Likes
Nemotron Mini 4B Instruct	4K / 8.4 GB	38413	186
Nemotron Mini 4B Instruct	4K / 16.7 GB	5	0
...otron 4 Mini Hindi 4B Instruct	4K / 16.7 GB	7	0
Nemotron 4 Mini Hindi 4B Base	4K / 16.7 GB	8	0
HindiNER 4B V1.0	4K / 8.4 GB	6	0
...4B Intermediate Gliner En Exp4	4K / 8.4 GB	129	0
...indi 4B Constraint Phase1 Exp1	4K / 8.4 GB	31	0
...4B Intermediate Gliner En Exp3	4K / 8.4 GB	6	0
...otron 4 Mini Hindi 4B Instruct	4K / 8.4 GB	13	0

Note: green Score (e.g. "73.2") means that the model is better than nvidia/Minitron-4B-Base.