Ollama VPS
Run small open-weight models on a CPU VPS with Ollama. Connect your applications to a private API and choose the models you download.
Start
$13.7/month
Ollama VPS
- 4 GB DDR4 Memory
- 2 vCPU High Frequency
- 50 GB NVMe SSD Storage
- 1 TB Bandwidth
- Ollama Latest · Ubuntu 24.04
Agency
$26.7/month
Ollama VPS
- 8 GB DDR4 Memory
- 4 vCPU High Frequency
- 50 GB NVMe SSD Storage
- 2 TB Bandwidth
- Ollama Latest · Ubuntu 24.04
Premium
$51.7/month
Ollama VPS
- 16 GB DDR4 Memory
- 8 vCPU High Frequency
- 50 GB NVMe SSD Storage
- 3 TB Bandwidth
- Ollama Latest · Ubuntu 24.04
Enterprise
$100.7/month
Ollama VPS
- 32 GB DDR4 Memory
- 16 vCPU High Frequency
- 50 GB NVMe SSD Storage
- 4 TB Bandwidth
- Ollama Latest · Ubuntu 24.04
Tools for your own server
Host an inference endpoint for small-model experiments and application development.
Small models on CPU
Start with a small quantized model for a prototype or a low-volume task. This VPS has no GPU, and inference speed depends on your model and CPU allocation.
An API for your applications
Use the Ollama API from your own tools through SSH forwarding. The installation provides the API service without an additional browser chat interface.
Models you choose
No model is downloaded automatically. Check each model's license and storage requirements before pulling it. Keep model files in persistent local storage.
Connect. Deploy. Grow - Globally
Start deploying your High Performance Cloud VPS worldwide. Reduce latency, with our Cloud VPS located near your users and equipped with local BGP access.
Silicon Valley
- Washington
Frankfurt
Istanbul
Riyadh
Dubai
Bangkok
Hanoi
Phnom Penh- Ho Chi Minh
Hong Kong
Taipei
Seoul
Johannesburg
Singapore
Manila
Dhaka
Sao Paulo- Jeddah
Tokyo
Cairo
Bahrain
Sofia
Athens
Kuala Lumpur
London
Muscat
Kuwait City
Marseille
Karachi
kathmandu
Moscow
Buenos Aires
Jakarta
Mexico City
Santiago
Yangon
Bogota
Baghdad
Doha
Lima
LightNode Worldwide Locations
0
21
Years Experience
150+
Data Centers
1,100,000+
Cloud Servers Launched
Choose a starting configuration
- OS
- Ubuntu 24.04, amd64.
- Resources
- Start with 2 vCPU, 4 GB RAM and 50 GB NVMe. Increase capacity for heavier workloads.
- Access
- Reach localhost:11434 through SSH forwarding. The local API has no application authentication. Keep access private.
- Costs
- MIT covers the application. VPS and paid external services cost separately.
Set up your application
- Choose this application, Latest, Ubuntu 24.04 and a plan with at least 4 GB RAM.
- Wait for installation, connect by SSH and read the access guide.
- Forward port 11434 over SSH and check the API. Pull a small model that fits your available memory.
- Send a test prompt, measure response time, and adjust the model or VPS size before connecting an agent.
Questions about this VPS
- First installation selects the current official application channel and records its container image. Repeated installation and restarts retain that image, configuration and data. Back up data and review upstream changes before a manual upgrade.
- This is a CPU-only installation. A 4 GB starting plan suits selected small models, not every model in the Ollama library. Model files, context, and concurrent requests all need memory. GPU acceleration is not included.
- Back up credentials and persistent data before upgrading. Follow upstream migration instructions. Restarting preserves the installed version and configuration.
Choose your application VPS
Select this application, Latest, Ubuntu 24.04 and a plan with at least 4 GB RAM.