How much VRAM does DBRX 132B (MoE) need?

DBRX 132B (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 DBRX 132B (MoE) on a consumer GPU?

DBRX 132B (MoE) is a 132B 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 DBRX 132B (MoE)'s context window?

DBRX 132B (MoE) supports a maximum context window of 32,768 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 DBRX 132B (MoE)?

For DBRX 132B (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 DBRX 132B (MoE) compare to similar models?

DBRX 132B (MoE) is a 132.0B MoE model from the Databricks family. Browse other Databricks 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.

DatabricksMoEDatabricks Open Model License

DBRX 132B (MoE)

132.0B

Parameters

36.0B

Active

32K

Max Context

MoE

Architecture

Mar 27, 2024

Released

Text

Modality

About DBRX 132B (MoE)

DBRX 132B is Databricks' MoE model — 132B total with 36B active across 16 experts (top-4 routing). One of the earliest major open MoE releases alongside Mixtral. At ~70 GB VRAM at Q4_K_M it requires server-class hardware. While newer MoE models have surpassed it in both quality and efficiency, DBRX remains historically significant and is still used in enterprise Databricks workflows. The Databricks Open Model License has some usage restrictions.

EnterpriseCodeResearch

Technical Specifications

Total Parameters132.0B

Active Parameters36.0B per token

ArchitectureMixture of Experts

Total Experts16

Active Experts4 per token

Attention TypeGQA (Grouped Query Attention)

Hidden Dimensiond = 6,144

Transformer Layers40

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	32K ctx
Q4_K_M0.50 B/W ~97% of FP16	68.38Datacenter GPU	73.23Datacenter GPU
Q8_01.00 B/W ~100% of FP16	136.6Cluster / Multi-GPU	141.5Cluster / Multi-GPU
F162.00 B/W Reference	273.1Cluster / Multi-GPU	277.9Cluster / Multi-GPU

Fits 24 GB consumer GPU

Fits 80 GB datacenter GPU

Requires cluster / multi-GPU

Model Transparency

Scores how well-documented and reproducible DBRX 132B (MoE) is across 12 provenance dimensions. Higher is more transparent.

Param Density80

Tokenizer78

Dataset58

Training66

Reproducibility72

Upstream74

Architecture76

Versioning68

Hardware76

Downstream68

License60

Identity80

85+ Transparent

70+ Adequate

<70 Opaque

Benchmark Scores

MMLU73.7

HumanEval65.2

GSM8K72

BBH76

IFEval68.5

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

Quick Reference

Total Params132.0B

Active Params36.0B

Context Window32K

ArchitectureMoE

Layers40

Hidden Dim6144

LicenseDatabricks Open Model License

VRAM Calculator

Select quantization and context to see exact VRAM requirements and compatible GPUs.

Find the right GPU for DBRX 132B (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