OWN · BRING YOUR OWN CLOUD

Run the platform in your cloud, under a platform licence

The platform runs in your cloud account, in the region you choose. Your provider bills the GPUs and the cloud infrastructure, and your datasets, checkpoints and weights stay in your account.

Your cloud accountyour region · your provider's bill
licensed
Your account
Nucleus platform
licensed
Training data
your buckets
Checkpoints and weights
your buckets
GPUs
billed by your provider
Nucleus billsone line on your invoice
from $50,000platform licence, per year
Managed fine-tuning jobs
included
The low-level training loop
included
Evaluations
included
Serving and weight export
included

In your account.

The platform runs in your cloud account, in the region you choose.

Your provider bills the GPUs.

GPU and cloud infrastructure are billed separately, by your provider, and the licence covers the platform.

Your data stays in your account.

Training datasets, checkpoints and finished weights stay inside your account and are never used to train anyone else's models.

Built for residency and air-gap.

Private deployment is built for data-residency and air-gapped requirements.

Private deployment

What a private deployment looks like

The training, evaluation and serving stack runs inside your cloud account, in the region you choose. Your applications call an OpenAI- or Anthropic-compatible endpoint inside that account, and your datasets, checkpoints and weights stay there. Your provider bills the GPUs the stack runs on.

Security and compliance
Results

Results from LLMs powered by Nucleus

18.3 days

Advance notice on a failure · Predictive Maintenance / Manufacturing

9

Systems on one thread · Real-Time Field Data Collection / Field engineering

6.2 hrs

Average time to resolution · Critical Incident Response / IT operations

USE CASES

Three workflows over internal systems

Manufacturing

Line 3's next failure, booked into the maintenance window

$177Kdowntime prevented this month
18.3 daysadvance notice on a failure
IT operations

Every incident past its four-hour SLA, escalated by name

8incidents past their four-hour SLA
6.2 hrsaverage time to resolution
Procurement

Six vendor contracts ranked before the renewal meeting

6vendors under review for Q4 renewal
3axes: cost, quality, response

Bring your own cloud, in four columns

Where it runs

  • Your cloud account
  • The region you choose
  • Your GPUs
  • Built for air-gapped requirements

What runs

  • Managed fine-tuning jobs
  • The low-level training loop
  • Evaluations
  • OpenAI + Anthropic compatible serving
  • Weight export, any checkpoint, any time

Your data

  • Datasets, checkpoints and weights stay in your account
  • Never used to train anyone else's models
  • AES-256 at rest, TLS in transit
  • Scoped API keys and role-based access
  • Append-only audit trail

The agreement

  • Platform licence from $50,000 / year
  • GPUs and cloud billed by your provider
  • Service Level Agreement and Support Policy
PRICING

What the licence costs

Platform licence

from $50,000/ year

Your GPU and cloud infrastructure are billed separately.

Pick your path

The rest of Own

Weight export

Any checkpoint, any format

Explore Weight export

Checkpoints

Kept until you delete them

Explore Checkpoints

Data residency

Choose your region

Explore Data residency

Storage

Metered by the GB-hour

Explore Storage

Run the platform in your cloud.

A platform licence from $50,000 a year, with the GPUs and the cloud billed by your provider.