Pricing

Pricing

Access 1,000s of NVIDIA GPUs with transparent, market-based pricing.

Compute

Compute

Instance type

GPUs

vCPUs

RAM

Disk

Spot (7-day avg)

Reserved (7-day avg)

NVIDIA A100 SXM4

1

24

225

GB

1,750

GB

$0.50

instance/hr

$0.88

instance/hr

NVIDIA A100 SXM4

2

48

450

GB

3,500

GB

$1.00

instance/hr

$1.76

instance/hr

NVIDIA A100 SXM4

4

96

900

GB

7,000

GB

$2.00

instance/hr

$3.54

instance/hr

NVIDIA A100 SXM4

8

192

1,800

GB

14,000

GB

$4.00

instance/hr

$7.14

instance/hr

NVIDIA H100 PCIe

8

104

1,007

GB

10,000

GB

$10.00

instance/hr

$10.92

instance/hr

NVIDIA H100 SXM5

8

192

1,814

GB

15,202

GB

$10.85

instance/hr

$12.03

instance/hr

NVIDIA H100 SXM5

8

104

1,800

GB

7,154

GB

$8.00

instance/hr

$12.00

instance/hr

Mithril's omnicloud platform sets compute prices algorithmically based on supply and demand. Prices shown here are updated weekly and represent an average of the last 7 days.

WEKA storage

InfiniBand interconnect

USA, Canada, Europe

API & batch job SDK

Batch inference

Batch inference

Model

Input price

Output price

Meta Llama 3.1 8B Instruct

$0.10

1M tokens

$0.10

1M tokens

Meta Llama 4 Scout 17B 16E Instruct

$0.10

1M tokens

$0.40

1M tokens

Meta Llama 4 Maverick 17B 128E Instruct FP8

$0.14

1M tokens

$0.43

1M tokens

Qwen 2.5 72B Instruct

$0.50

1M tokens

$0.50

1M tokens

Simple API

Open or custom models

Token-based pricing

24h batch job completion

Meta Llama 3.1 8B Instruct

$0.10

Input

$0.10

Output

Meta Llama 4 Scout 17B 16E Instruct

$0.10

Input

$0.40

Output

Meta Llama 4 Maverick 17B 128E Instruct FP8

$0.14

Input

$0.43

Output

Qwen 2.5 72B Instruct

$0.50

Input

$0.50

Output

Per 1M tokens

Storage

Storage

Type

Device

Rate

File share

NVMe SSD

$0.08

GB/month

GB/mo

Block

NVMe SSD

$0.08

GB/month

GB/mo

WEKA storage

No ingress/egress

File share & block

Ultra-fast