Pricing
Simple plans that grow with your models
Model design, analysis, supported dataset profiling, native Training and supported LoRA Fine-tuning run on your selected machine. If you use the optional Agent, its configured provider receives the prompt and project design/data profile; provider usage is billed separately.
Community
For anyone exploring real model architectures and local compute.
Free
Free forever
- Model architecture analysis
- Import from Hugging Face
- Basic compute estimation
- Limited hardware comparisons
- 3 projects
- Basic reports
Developer
RecommendedFor independent developers, researchers and small teams building with their own models.
$4.99/ month
Indicative monthly price
- Unlimited projects
- Full cost estimation
- Hardware comparison
- NEURAX assistant, on your own key
- Import from Hugging Face
- Training-time estimation
- Hyperparameter impact analysis
- Training scenarios
- Exportable reports
- Local, private workspaces
Pro
For teams whose compute bill is a line item.
$20/ month
Indicative monthly price
Everything in Developer, plus
- Advanced stress testing
- Multi-year cost projections
- Batch experimentation
- Advanced optimization analysis
- In-depth reports
- Priority updates
Enterprise
For organisations standardising how AI compute is measured.
Custom
Priced for your organisation
Everything in Pro, plus
- Volume licensing and invoicing
- The assistant on your internal model endpoint
- Onboarding for your teams
- Help with your security review
- A direct line to the people building NEURAX
Prices shown are monthly USD guides. Paid plan subscriptions are not open yet; availability and terms are agreed directly before any purchase.
Compare the plans
| Feature | Community | Developer | Pro | Enterprise |
|---|---|---|---|---|
| Projects | 3 | Unlimited | Unlimited | Unlimited |
| Architecture analysis | Included | Included | Included | Included |
| Import from Hugging Face | Included | Included | Included | Included |
| Compute estimation | Basic | Full | Full | Full |
| Hardware comparison | Limited | Included | Included | Included |
| NEURAX assistant | Not included | Included | Included | Included |
| Training and Fine-tuning estimation | Not included | Included | Included | Included |
| Hyperparameter impact analysis | Not included | Included | Included | Included |
| Training and Fine-tuning scenarios | Not included | Included | Included | Included |
| Reports | Basic | Exportable | In-depth | In-depth |
| Stress testing | Not included | Not included | Included | Included |
| Multi-year cost projections | Not included | Not included | Included | Included |
| Batch experimentation | Not included | Not included | Included | Included |
| Advanced optimization analysis | Not included | Not included | Included | Included |
| Volume licensing and onboarding | Not included | Not included | Not included | Included |
About paying for NEURAX
What leaves my machine when I use NEURAX?
Supported dataset files and training stay local. A public Hugging Face or ModelScope import requests model configuration online. If you use the optional Agent, your configured provider receives the prompt and project design/data profile; NEURAX says raw dataset contents and the local path are not sent. The provider's own terms, logging and charges apply.
Is the assistant's model usage included?
No, and that is on purpose. The assistant uses your own key with your own provider, or an endpoint inside your network, so nobody sits between you and it.
Can I start on Community and upgrade later?
That is the intended upgrade path. Community is available now; paid subscriptions are not open yet. Your projects stay on your machine, and you can ask us about a paid plan through a demo request.
What does yearly billing change?
Paid subscriptions are not open yet. We will confirm the available billing periods and exact terms directly before any purchase.
Request a demo
See NEURAX on your own models
Tell us what you train and where it runs. We will walk you through NEURAX on an architecture like yours: memory, time and cost on the hardware you actually have.