CPU S
A private Ollama endpoint for small models
€11.90/mo
Billed annually · €142.80
- 2 vCPU · 4 GB RAM · CPU inference
- 20 GB storage
- Models up to about 3B parameters
- Model library and pulls
- Inference test console
AI for your workflow. Your own endpoint.
Run supported models through an Ollama endpoint with CPU resources. Choose a plan around the model size and expected use.
Choose the room your business needs.
Compare the resources, choose your billing period, and continue with your selected plan.
Most popular
CPU M
For larger models and steadier throughput
€23.90/mo
Billed annually · €286.80
- 4 vCPU · 8 GB RAM · CPU inference
- 40 GB storage
- Models up to about 8B parameters
- Auto-sleep when idle
- Priority support
Prices in EUR, VAT not included. Annual prices above show the monthly equivalent.
Built around the way your business works.
A model endpoint for your tools
Your application sends a request to Ollama and receives a model response. The hosting provides the resources that run the model.
- An Ollama endpoint
- A library for model selection and download
- A console for trial inference requests
Choose a model that fits
Model size affects memory needs. CPU S and CPU M list different memory and storage limits with approximate model-size guidance.
- Compare the model with the plan memory
- Allow storage for downloaded models
- Treat model-size guidance as an estimate
Know the capacity you buy
These offers run inference on a CPU. Inference means generating a response from an existing model. A GPU is not part of these plans.
- CPU-based model execution
- Published memory and storage limits
- Monthly or annual billing for the selected plan
Questions, answered
Do the plans include a GPU?
No. CPU S and CPU M use CPU resources. Their published specifications do not include a GPU.
Is model training included?
The plans describe running existing models through Ollama. Model training is not part of the listed features.
Will every request have the same response time?
Response time depends on the model, request and workload. The resource plan does not promise a fixed response time.
Your next step is a clear one.
Choose your resources and continue to your account. Your plan and billing period travel with you.