What are the hardware requirements for WizardLM 2 8x22B?

WizardLM 2 8x22B requires approximately 216.8 GB of VRAM and supports a context window of 64K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed WizardLM 2 8x22B and how large is it?

WizardLM 2 8x22B is developed by alpindale, a model with 140.6b 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 WizardLM 2 8x22B?

WizardLM 2 8x22B is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

WizardLM 2 8x22B by alpindale — VRAM 216.8GB, 64K context

Name: WizardLM 2 8x22B
Rating: 2.5 (2 reviews)
Author: alpindale

WizardLM 2 8x22B is an open-source language model by alpindale. Features: 140.6b LLM, VRAM: 216.8GB, Context: 64K, License: apache-2.0, MoE, LLM Explorer Score: 0.26.

Arxiv:2304.12244 Arxiv:2306.08568 Arxiv:2308.09583 Conversational Deploy:azure Endpoints compatible Mixtral Model-index Moe Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/alpindale/WizardLM-2-8x22B

WizardLM 2 8x22B Parameters and Internals

Model Type

Mixture of Experts (MoE), text-generation

Supported Languages

Multilingual (yes)

Training Details

Model Architecture:

Mixture of Experts (MoE)

Input Output

Input Format:

multi-turn conversation prompts

LLM Name	WizardLM 2 8x22B
Repository 🤗	https://huggingface.co/alpindale/WizardLM-2-8x22B
Model Size	140.6b
Required VRAM	216.8 GB
Updated	2026-08-01
Maintainer	alpindale
Model Type	mixtral
Model Files	5.0 GB: 1-of-59 4.8 GB: 2-of-59 4.8 GB: 3-of-59 4.8 GB: 4-of-59 4.8 GB: 5-of-59 4.8 GB: 6-of-59 4.8 GB: 7-of-59 4.8 GB: 8-of-59 4.8 GB: 9-of-59 4.8 GB: 10-of-59 4.8 GB: 11-of-59 4.8 GB: 12-of-59 4.8 GB: 13-of-59 4.8 GB: 14-of-59 4.8 GB: 15-of-59 4.8 GB: 16-of-59 4.8 GB: 17-of-59 4.8 GB: 18-of-59 4.8 GB: 19-of-59 4.8 GB: 20-of-59 4.8 GB: 21-of-59 4.8 GB: 22-of-59 4.8 GB: 23-of-59 4.9 GB: 24-of-59 5.0 GB: 25-of-59 5.0 GB: 26-of-59 4.9 GB: 27-of-59 4.8 GB: 28-of-59 4.8 GB: 29-of-59 4.8 GB: 30-of-59 4.8 GB: 31-of-59 4.8 GB: 32-of-59 4.8 GB: 33-of-59 4.8 GB: 34-of-59 4.8 GB: 35-of-59 4.8 GB: 36-of-59 4.8 GB: 37-of-59 4.8 GB: 38-of-59 4.8 GB: 39-of-59 4.8 GB: 40-of-59 4.8 GB: 41-of-59 4.8 GB: 42-of-59 4.8 GB: 43-of-59 4.8 GB: 44-of-59 4.8 GB: 45-of-59
Model Architecture	MixtralForCausalLM
License	apache-2.0
Context Length	65536
Model Max Length	65536
Transformers Version	4.36.2
Vocabulary Size	32000
Torch Data Type	bfloat16

Rank the WizardLM 2 8x22B 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 WizardLM 2 8x22B

Best Alternatives	Context / RAM	Downloads	Likes
Mixtral 8x22B Instruct V0.1	64K / 221.4 GB	71589	756
Zephyr Orpo 141B A35b V0.1	64K / 207.2 GB	66	270
Mixtral 8x22B V0.1	64K / 221.6 GB	5069	239
...ixtral 8x22B Instruct V0.1 FP8	64K / 140.9 GB	11259	0
Mixtral 8x22B V0.3	64K / 221.4 GB	11	3
Mixtral 8x22B V0.1	64K / 212 GB	109	671
XLAM 8x22b R	64K / 211.8 GB	15934	45
Dolphin 2.9.2 Mixtral 8x22b	64K / 207.2 GB	9311	40
Dolphin 2.9.2 Mixtral 8x22b	64K / 207.2 GB	8402	43
...igHuggyD Grey WizardLM 2 8x22B	64K / 216.6 GB	21	4

Note: green Score (e.g. "73.2") means that the model is better than alpindale/WizardLM-2-8x22B.