Updated minutes ago
Databricks

DBRX Instruct

Databricks · moe · 132B parameters · 32,768 context

Quality
50.0
Calculate ROI →Compare with othersFine-Tune This Model →

Parameters

132B

Context Window

32K tokens

Architecture

MoE

Best GPU

B200 SXM

Cheapest API

$1.20/M

Intelligence Brief

DBRX Instruct is a 132B parameter Mixture-of-Experts (16 experts, 4 active) model from Databricks, featuring Grouped Query Attention (GQA) with 40 layers and 6,144 hidden dimensions. With a 32,768 token context window, it supports tools, structured output, code, math. The most cost-effective API deployment is via together at $1.20/M output tokens. For self-hosted inference, B200 SXM delivers optimal throughput at $4261/month.

Architecture Details

TypeMOE
Total Parameters132B
Active Parameters36B
Layers40
Hidden Dimension6,144
Attention Heads48
KV Heads8
Head Dimension128
Vocab Size100,352
Total Experts16
Active Experts4

Memory Requirements

BF16 Weights

264.0 GB

FP8 Weights

132.0 GB

INT4 Weights

66.0 GB

KV-Cache per Token163840 bytes
Activation Estimate2.00 GB

Fits on (single GPU) — most practical first

BEST FITH100 SXM80GBINT4H100 PCIe80GBINT4A100 80GB SXM80GBINT4A100 80GB PCIe80GBINT4H100 NVL94GBINT4H2096GBINT4Gaudi 296GBINT4GH20096GBINT4Trainium296GBINT4Instinct MI250X128GBINT4Gaudi 3128GBINT4Gaudi 3 HL-325L128GBINT4

GPU Compatibility Matrix

DBRX Instruct is compatible with 20% of GPU configurations across 41 GPUs at 3 precision levels.

BF16 (Full)
FP8 (Half)
INT4 (Quarter)
Blackwell(7 GPUs)
B200 NVL (pair)360GB
27%
63%
82%
B300288GB
8%
54%
77%
B100 SXM192GB
✗
31%
66%
GB200 NVL72 (per GPU)192GB
✗
31%
66%
Hopper(7 GPUs)
H100 NVL 94GB (per GPU pair)188GB
✗
30%
65%
H200 SXM141GB
✗
6%
53%
H2096GB
✗
✗
31%
GH20096GB
✗
✗
31%
Ada Lovelace(11 GPUs)
L40S48GB
✗
✗
✗
L4048GB
✗
✗
✗
RTX 6000 Ada48GB
✗
✗
✗
L2048GB
✗
✗
✗
Ampere(16 GPUs)
A100 80GB SXM80GB
✗
✗
18%
A100 80GB PCIe80GB
✗
✗
18%
A1664GB
✗
✗
✗
RTX A600048GB
✗
✗
✗
Legend:No fitVery tightTightModerateGoodExcellent

GPU Recommendations

B200 SXMoptimal

FP8 · 1 GPU · tensorrt-llm

100/100

score

Throughput

280.0 tok/s

Latency (ITL)

3.6ms

Est. TTFT

1ms

Cost/Month

$4261

Cost/M Tokens

$5.79

Use this config →
B100 SXMoptimal

FP8 · 1 GPU · tensorrt-llm

100/100

score

Throughput

280.0 tok/s

Latency (ITL)

3.6ms

Est. TTFT

1ms

Cost/Month

$4271

Cost/M Tokens

$5.80

Use this config →
GB200 NVL72 (per GPU)optimal

FP8 · 1 GPU · tensorrt-llm

100/100

score

Throughput

280.0 tok/s

Latency (ITL)

3.6ms

Est. TTFT

1ms

Cost/Month

$6169

Cost/M Tokens

$8.38

Use this config →

Deployment Options

API

API Deployment

together

$1.20/M

output tokens

Self-Hosted

Single GPU

B200 SXM

$4261/mo

Min VRAM: 132 GB

Scale

Multi-GPU

H200 SXM x2

280.0 tok/s

TP· $5106/mo

Open Full Calculator →

API Pricing Comparison

ProviderInput $/MOutput $/MBadges
together$1.20$1.20
Cheapest
databricks$0.75$2.25
Low Input

Cost Analysis

ProviderInput $/MOutput $/M~Monthly Cost
togetherBest Value$1.20$1.20$12
databricks$0.75$2.25$15

Cost per 1,000 Requests

Short (500 tok)

$0.84

via together

Medium (2K tok)

$3.36

via together

Long (8K tok)

$12.00

via together

Performance Estimates

Throughput by GPU

B200 SXM
280.0 tok/s
B100 SXM
280.0 tok/s
GB200 NVL72 (per GPU)
280.0 tok/s

VRAM Breakdown (B200 SXM, FP8)

Weights
Weights 132.0 GBKV-Cache 1.3 GBActivations 16.0 GBOverhead 6.6 GB

Precision Impact

bf16

264.0 GB

weights/GPU

fp8

132.0 GB

weights/GPU

~280.0 tok/s

int4

66.0 GB

weights/GPU

Capabilities

Features

✓ Tool Use✗ Vision✓ Code✓ Math✗ Reasoning✗ Multilingual✓ Structured Output

Supported Frameworks

vllmsglangtgitensorrt-llm

Supported Precisions

BF16 (default)FP8INT4

Where to Deploy DBRX Instruct

Similar Models

DBRX Base

132B params · moe

Quality: 50

from $2.25/M

More expensiveCompare →

Mistral Large 2411

123B params · dense

Quality: 75

from $6.00/M

Larger context, Higher quality, More expensiveCompare →

Mistral Large 2

123B params · dense

Quality: 75

from $2.50/M

Larger context, Higher quality, More expensiveCompare →

Mixtral 8x22B

141B params · moe

Quality: 65

from $1.20/M

Larger context, Higher qualityCompare →

Nemotron-3 Super 120B

120B params · dense

Quality: 84

from $2.40/M

Larger context, Higher quality, More expensiveCompare →

Frequently Asked Questions

How much VRAM does DBRX Instruct need for inference?

DBRX Instruct requires approximately 264.0 GB of VRAM at BF16 precision, 132.0 GB at FP8, or 66.0 GB at INT4 quantization. Additional VRAM is needed for KV-cache (163840 bytes per token) and activations (~2.00 GB).

What is the best GPU for DBRX Instruct?

The top recommended GPU for DBRX Instruct is the B200 SXM using FP8 precision. It achieves approximately 280.0 tokens/sec at an estimated cost of $4261/month ($5.79/M tokens). Score: 100/100.

How much does DBRX Instruct inference cost?

DBRX Instruct API inference starts from $1.20/M input tokens and $1.20/M output tokens. Self-hosted inference costs depend on your GPU configuration — use our ROI calculator for a detailed breakdown.