How much VRAM does Mixtral 8x22B (MoE) need?

Mixtral 8x22B (MoE) requires approximately 0.0 GB VRAM at Q4_K_M quantization with 1K context and 10% overhead. At Q8_0 it needs ~0.0 GB, and at FP16 ~0.0 GB. Use the PCPARTGUIDE VRAM Calculator for exact requirements at any context length and overhead setting.

Can I run Mixtral 8x22B (MoE) on a consumer GPU?

Mixtral 8x22B (MoE) is a 141B parameter model and requires datacenter or multi-GPU setups. At Q4_K_M, it needs approximately 0.0 GB VRAM, which exceeds consumer GPU limits. It can run on H100 (80 GB) or A100 (80 GB) datacenter GPUs, or across multiple consumer GPUs with tensor parallelism.

What is Mixtral 8x22B (MoE)'s context window?

Mixtral 8x22B (MoE) supports a maximum context window of 65,536 tokens. The default context is 4,096 tokens. Larger contexts require proportionally more VRAM for the KV cache — use the VRAM Calculator to estimate exact requirements at your target context length.

What quantization level should I use for Mixtral 8x22B (MoE)?

For Mixtral 8x22B (MoE), start with Q4_K_M (~0.0 GB) if VRAM-limited. Q8_0 offers better quality at ~2× the VRAM. If you have enough VRAM, Q6_K (~1.5× Q4 VRAM) provides a good balance. For models this large, Q3_K_M may be necessary to fit on consumer hardware.

How does Mixtral 8x22B (MoE) compare to similar models?

Mixtral 8x22B (MoE) is a 141.0B MoE model from the Mistral family. Browse other Mistral models on the PCPARTGUIDE directory for comparable alternatives with different parameter counts and VRAM requirements. Use the Model Transparency scores to assess documentation quality and reproducibility.

MistralMoEApache 2.0

Mixtral 8x22B (MoE)

141.0B

Parameters

39.0B

Active

64K

Max Context

MoE

Architecture

Apr 17, 2024

Released

Text

Modality

About Mixtral 8x22B (MoE)

Mixtral 8x22B is the larger sibling of Mixtral 8x7B — 141B total parameters with 39B active per token. It delivers quality competitive with dense 70B models but requires significant VRAM (~75 GB at Q4_K_M). The Apache 2.0 license makes it attractive for enterprise deployments, though most users will need server-class hardware or aggressive quantization. Supports 65K context.

EnterpriseCodeMultilingualCommercial

Technical Specifications

Total Parameters141.0B

Active Parameters39.0B per token

ArchitectureMixture of Experts

Total Experts8

Active Experts2 per token

Attention TypeGQA (Grouped Query Attention)

Hidden Dimensiond = 6,144

Transformer Layers56

Attention Heads48

KV Headsn_kv = 8

Head Dimensiond_head = 128

Activation FunctionSwiGLU

NormalizationRMSNorm

Position EmbeddingRoPE

System Requirements

Estimated VRAM at 10% overhead for different quantization methods and context sizes.

Quantization	1K ctx	64K ctx
Q4_K_M0.50 B/W ~97% of FP16	73.10Datacenter GPU	86.88Cluster / Multi-GPU
Q8_01.00 B/W ~100% of FP16	146.0Cluster / Multi-GPU	159.8Cluster / Multi-GPU
F162.00 B/W Reference	291.7Cluster / Multi-GPU	305.5Cluster / Multi-GPU

Fits 24 GB consumer GPU

Fits 80 GB datacenter GPU

Requires cluster / multi-GPU

Model Transparency

Scores how well-documented and reproducible Mixtral 8x22B (MoE) is across 12 provenance dimensions. Higher is more transparent.

Param Density86

Tokenizer82

Dataset58

Training68

Reproducibility76

Upstream80

Architecture82

Versioning74

Hardware80

Downstream86

License98

Identity84

85+ Transparent

70+ Adequate

<70 Opaque

Benchmark Scores

MMLU77.8

HumanEval67.1

GSM8K83.7

BBH82

IFEval73.4

Arena Elo1180

Sources: Open LLM Leaderboard, LMSYS Chatbot Arena, model technical reports.

Quick Reference

Total Params141.0B

Active Params39.0B

Context Window64K

ArchitectureMoE

Layers56

Hidden Dim6144

LicenseApache 2.0

VRAM Calculator

Layers

Context

256K

41.0B active

Find the right GPU for Mixtral 8x22B (MoE)

Use the interactive VRAM Calculator to see exactly how much memory you need at any quantization level, context length, and overhead setting.

Open VRAM Calculator Best GPU Guide VRAM Guide