What are the hardware requirements for Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw?

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw requires approximately 70.3 GB of VRAM and supports a context window of 8K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw and how large is it?

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw is developed by gbueno86, a model with 120b 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 Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw?

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw by gbueno86 — VRAM 70.3GB, 8K context

Name: Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw
Author: gbueno86

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw is an open-source language model by gbueno86. Features: 120b LLM, VRAM: 70.3GB, Context: 8K, Quantized, Instruction-Based, Merged, LLM Explorer Score: 0.12.

Merged Model Conversational Endpoints compatible Exl2 Instruct Llama Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/gbueno86/Meta-Llama-3-Instruct-120b-Cat-a-llama-exl2-4.5bpw

Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw Parameters and Internals

Additional Notes

This model is a result of merging two pre-trained language models using the passthrough merge method. It demonstrates unexpected intelligence and capability, as observed by the user.

Training Details

Methodology:

Passthrough merge method, combining layers from several models using mergekit.

LLM Name	Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw
Repository 🤗	https://huggingface.co/gbueno86/Meta-Llama-3-Instruct-120b-Cat-a-llama-exl2-4.5bpw
Base Model(s)	...ma 3 Instruct 120B Cat A Llama gbueno86/Meta-Llama-3-Instruct-120b-Cat-a-llama
Merged Model	Yes
Model Size	120b
Required VRAM	70.3 GB
Updated	2026-07-06
Maintainer	gbueno86
Model Type	llama
Instruction-Based	Yes
Model Files	8.6 GB: 1-of-9 8.6 GB: 2-of-9 8.5 GB: 3-of-9 8.5 GB: 4-of-9 8.4 GB: 5-of-9 8.5 GB: 6-of-9 8.5 GB: 7-of-9 8.5 GB: 8-of-9 2.2 GB: 9-of-9
Quantization Type	exl2
Model Architecture	LlamaForCausalLM
Context Length	8192
Model Max Length	8192
Transformers Version	4.40.1
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw 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 Meta Llama 3 Instruct 120B Cat A Llama EXL2 4.5bpw

Best Alternatives	Context / RAM	Downloads	Likes
...t 120B Cat A Llama EXL2 5.5bpw	8K / 85.3 GB	5	0
...egaDolphin 120B 2.9bpw H6 EXL2	4K / 44.3 GB	2	3
...gaDolphin 120B 2.65bpw H6 EXL2	4K / 40.5 GB	3	2
...egaDolphin 120B 4.0bpw H6 EXL2	4K / 60.8 GB	4	1
...ma 3 Instruct 120B Cat A Llama	8K / 243.9 GB	2	1
Meta Llama 3 225B Instruct	8K / 443.2 GB	274	18
...0B Instruct Abliterated Merged	8K / 243.7 GB	1	1
MegaDolphin 120B GPTQ	4K / 61.1 GB	25	4
MegaDolphin 120B AWQ	4K / 63.3 GB	7	2
Koishi 120B Qlora Gptq	4K / 59.8 GB	4	1

Note: green Score (e.g. "73.2") means that the model is better than gbueno86/Meta-Llama-3-Instruct-120b-Cat-a-llama-exl2-4.5bpw.