Skip to content
Updated minutes ago
Mistral

Mistral Nemo 12B

Mistral AI · dense · 12B parameters · 131,072 context

Quality
62.0

Parameters

12B

Context Window

128K tokens

Architecture

Dense

Best GPU

A100 40GB SXM

Cheapest API

$0.13/M

Quality Score

62/100

Intelligence Brief

Mistral Nemo 12B is a 12B parameter DENSE model from Mistral AI, featuring Grouped Query Attention (GQA) with 40 layers and 5,120 hidden dimensions. With a 131,072 token context window, it supports tools, structured output, code, math, multilingual. On standardized benchmarks, it achieves MMLU 68, HumanEval 38, GSM8K 65. The most cost-effective API deployment is via deepinfra at $0.13/M output tokens. For self-hosted inference, A100 40GB SXM delivers optimal throughput at $807/month.

Architecture Details

TypeDENSE
Total Parameters12B
Active Parameters12B
Layers40
Hidden Dimension5,120
Attention Heads32
KV Heads8
Head Dimension128
Vocab Size131,072

Memory Requirements

BF16 Weights

24.0 GB

FP8 Weights

12.0 GB

INT4 Weights

6.0 GB

KV-Cache per Token163840 bytes
Activation Estimate1.50 GB

GPU Compatibility Matrix

Mistral Nemo 12B is compatible with 82% of GPU configurations across 41 GPUs at 3 precision levels.

BF16 (Full)
FP8 (Half)
INT4 (Quarter)
Blackwell(7 GPUs)
B200 NVL (pair)360GB
B300288GB
B100 SXM192GB
GB200 NVL72 (per GPU)192GB
Hopper(7 GPUs)
H100 NVL 94GB (per GPU pair)188GB
H200 SXM141GB
H2096GB
GH20096GB
Ada Lovelace(11 GPUs)
L40S48GB
L4048GB
RTX 6000 Ada48GB
L2048GB
Ampere(16 GPUs)
A100 80GB SXM80GB
A100 80GB PCIe80GB
A1664GB
RTX A600048GB
Legend:No fitVery tightTightModerateGoodExcellent

GPU Recommendations

A100 40GB SXMoptimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

349.9 tok/s

Latency (ITL)

2.9ms

Est. TTFT

0ms

Cost/Month

$807

Cost/M Tokens

$0.88

Use this config →
RTX A6000optimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

172.8 tok/s

Latency (ITL)

5.8ms

Est. TTFT

1ms

Cost/Month

$465

Cost/M Tokens

$1.02

Use this config →
A40optimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

156.6 tok/s

Latency (ITL)

6.4ms

Est. TTFT

1ms

Cost/Month

$399

Cost/M Tokens

$0.97

Use this config →

Deployment Options

API

API Deployment

deepinfra

$0.13/M

output tokens

Self-Hosted

Single GPU

A100 40GB SXM

$807/mo

Min VRAM: 12 GB

Scale

Multi-GPU

RTX 3090 x2

343.7 tok/s

TP· $361/mo

API Pricing Comparison

ProviderInput $/MOutput $/MBadges
deepinfra$0.13$0.13
Cheapest
together$0.18$0.18

Cost Analysis

ProviderInput $/MOutput $/M~Monthly Cost
deepinfraBest Value$0.13$0.13$1
together$0.18$0.18$2

Cost per 1,000 Requests

Short (500 tok)

$0.09

via deepinfra

Medium (2K tok)

$0.36

via deepinfra

Long (8K tok)

$1.30

via deepinfra

Performance Estimates

Throughput by GPU

A100 40GB SXM
349.9 tok/s
RTX A6000
172.8 tok/s
A40
156.6 tok/s

VRAM Breakdown (A100 40GB SXM, BF16)

Weights
Act
Weights 24.0 GBKV-Cache 2.7 GBActivations 12.0 GBOverhead 1.9 GB

Precision Impact

bf16

24.0 GB

weights/GPU

~349.9 tok/s

fp8

12.0 GB

weights/GPU

int4

6.0 GB

weights/GPU

Quality Benchmarks

Average
64th percentile across all models
MMLU
68.0
Below Average (27th pctile)
HumanEval
38.0
Bottom 25% (23th pctile)
GSM8K
65.0
Below Average (26th pctile)
MT-Bench
75.0
Bottom 25% (0th pctile)

Capabilities

Features

Tool Use Vision Code Math Reasoning Multilingual Structured Output

Supported Frameworks

vllmsglangtgitensorrt-llmollama

Supported Precisions

BF16 (default)FP8INT4

Where to Deploy Mistral Nemo 12B

Similar Models

Larger context, Lower quality, More expensiveCompare →

Gemma 3 12B

12B params · dense

Quality: 71

from $0.10/M

Higher qualityCompare →

Pixtral 12B

12B params · dense

Quality: 50

from $0.15/M

Lower qualityCompare →

FLUX.1 Dev

12B params · dense

Quality: 50

from $25.00/M

Smaller context, Lower quality, More expensiveCompare →
Smaller context, Lower qualityCompare →

Frequently Asked Questions

How much VRAM does Mistral Nemo 12B need for inference?

Mistral Nemo 12B requires approximately 24.0 GB of VRAM at BF16 precision, 12.0 GB at FP8, or 6.0 GB at INT4 quantization. Additional VRAM is needed for KV-cache (163840 bytes per token) and activations (~1.50 GB).

What is the best GPU for Mistral Nemo 12B?

The top recommended GPU for Mistral Nemo 12B is the A100 40GB SXM using BF16 precision. It achieves approximately 349.9 tokens/sec at an estimated cost of $807/month ($0.88/M tokens). Score: 95/100.

How much does Mistral Nemo 12B inference cost?

Mistral Nemo 12B API inference starts from $0.13/M input tokens and $0.13/M output tokens. Self-hosted inference costs depend on your GPU configuration — use our ROI calculator for a detailed breakdown.