GLWS · Model runtimes
Run supported language models locally and connect applications to a model runtime.
Common uses of Ollama.
Model size strongly affects memory and response time. CPU-based VPS plans do not include a GPU. Restrict access to model endpoints.
Planning examples, not official minimum requirements.
Model runtimes
Check the current supported operating systems, deployment methods, dependencies and licence terms. What you need depends on your workload, not just the application's name.
Large datasets and busy workloads can need far more memory, storage or different hardware. Any external API or service subscriptions are separate.
Read the Ollama documentation ↗ (opens in a new tab)Advice specific to this application.
Select a model only after reviewing its memory and storage requirements. Larger model files and parallel requests can change the capacity needed substantially. The CPU-only VPS examples here do not promise GPU acceleration or interactive performance for any specific model.
Identify which applications need to reach the runtime. Keep access restricted rather than exposing an unrestricted endpoint to the public network. Account for model storage and a repeatable way to recreate the configuration; model files and application data may need different backup treatment.
Linux guides that help with this setup.
Install Ollama, pull a small quantised model, measure its real speed on your plan and keep the API private to the server.
Read guide →Work out how much memory, disk and CPU a quantised model needs, then test on a small plan and measure before you upgrade.
Read guide →Keep AI services off the open internet: bind to localhost, avoid the Docker firewall bypass, add authentication and verify from outside.
Read guide →Four steps that make the environment easier to maintain.
Read the publisher's instructions and choose a supported installation method.
Protect the dashboard, credentials and network endpoints before you expose any service.
Work out where the application keeps its state and store it separately from replaceable runtime files.
Back up the data and configuration, then check that you can restore them.
Run through it before and after you install.
$ application-checklist --name ollama
[ ] Supported deployment method
[ ] Authentication and access restrictions
[ ] Independent data backup
[ ] Updates and restoration tested
root@glws:~#
Quick answers before you choose resources.
No. The configurator only saves your application preference. It does not connect to any infrastructure or run an installer.
Use the planning example as a starting point only. Check the publisher's requirements and measure the whole stack under a realistic workload.
You do. Self-hosted applications need updates, configuration and recovery planning from the operator. The publisher's documentation describes supported maintenance.
The knowledge base covers Linux access, packages, services, firewalls and backups.
Other tools you might run alongside or instead.
Website, email, DNS and database management in a hosting dashboard.
Explore application →A self-hosted assistant for connecting model-powered tasks with tools and messaging.
Explore application →An agent workspace for conversations, tools and automation on infrastructure you control.
Explore application →Connect services and build repeatable workflows with a self-hosted automation platform.
Explore application →