What are the hardware requirements for Llama 3 Aplite Instruct 4x8B MoE?

Llama 3 Aplite Instruct 4x8B MoE requires approximately 50 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 Llama 3 Aplite Instruct 4x8B MoE and how large is it?

Llama 3 Aplite Instruct 4x8B MoE is developed by raincandy-u, a model with 24.9b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Llama 3 Aplite Instruct 4x8B MoE perform on standard benchmarks?

Llama 3 Aplite Instruct 4x8B MoE has the following published scores: MMLU 67.47. Compare against reference models on this page or on the LLM Explorer leaderboards.

Llama 3 Aplite Instruct 4x8B MoE by raincandy-u — VRAM 50GB, 8K context

Name: Llama 3 Aplite Instruct 4x8B MoE
Author: raincandy-u

Llama 3 Aplite Instruct 4x8B MoE is an open-source language model by raincandy-u. Features: 24.9b LLM, VRAM: 50GB, Context: 8K, License: other, MoE, Instruction-Based, LLM Explorer Score: 0.13, Arc: 61.9, HellaSwag: 81.1, MMLU: 67.5, GSM8K: 61.9.

Base model:finetune:meta-llama... Base model:meta-llama/meta-lla... Code Conversational En Endpoints compatible Facebook Instruct Llama Llama-3 Meta Mixtral Model-index Moe Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/raincandy-u/Llama-3-Aplite-Instruct-4x8B-MoE

Llama 3 Aplite Instruct 4x8B MoE Parameters and Internals

Model Type

text-generation

Additional Notes

This model is a research experiment and may generate incorrect or harmful content. Outputs should not be taken as factual or representative of the views of the model's creator or any individual.

Training Details

Data Sources:

Meta-Llama-3-8B-Instruct, Llama3-8B-OpenHermes-DPO, Llama-3-SLERP-8B, hf-llama3-8b-orpo-v0.0

LLM Name	Llama 3 Aplite Instruct 4x8B MoE
Repository 🤗	https://huggingface.co/raincandy-u/Llama-3-Aplite-Instruct-4x8B-MoE
Base Model(s)	Meta Llama 3 8B Instruct meta-llama/Meta-Llama-3-8B-Instruct
Model Size	24.9b
Required VRAM	50 GB
Updated	2026-07-30
Maintainer	raincandy-u
Model Type	mixtral
Instruction-Based	Yes
Model Files	9.9 GB: 1-of-6 10.0 GB: 2-of-6 10.0 GB: 3-of-6 9.9 GB: 4-of-6 9.1 GB: 5-of-6 1.1 GB: 6-of-6
Supported Languages	en
Model Architecture	MixtralForCausalLM
License	other
Context Length	8192
Model Max Length	8192
Transformers Version	4.39.3
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|begin_of_text\|>
Vocabulary Size	128256
Torch Data Type	float16

Rank the Llama 3 Aplite Instruct 4x8B MoE 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 Aplite Instruct 4x8B MoE

Best Alternatives	Context / RAM	Downloads	Likes
L3.1 MoE 4x8B V0.2	128K / 50.1 GB	4	2
OpenCrystal V4 L3 4x8B	8K / 50 GB	5	2
L3 Arcania 4x8b	8K / 50 GB	42	29
Llama 3 8B Instruct MoE 2	8K / 50 GB	6	0
Llama 3 4x8B	8K / 49.9 GB	11	0

Note: green Score (e.g. "73.2") means that the model is better than raincandy-u/Llama-3-Aplite-Instruct-4x8B-MoE.