Does Ollama run without a GPU?

Yes, but fitness and interactivity depend on the model and resources.

What should be clarified before installation?

Use case, user group, model, resources, network and operating ownership.

First define which tasks the model should support, who may use it and whether it is exposed through a user interface or API.

Review memory, CPU/GPU, operating system and drivers. Ollama documentation lists supported GPU paths; practical fit depends on the intended model.

Select models by task, language, licence and resource limits. Restrict network access and document model versions.

Protect configuration and operating knowledge. Plan how updates, testing and questions after installation are handled.

Preparation	target state, user group, hardware and ownership
Set-up	runtime, model storage, access and baseline configuration
Operations	updates, review and documented changes

Get technical context

We clarify requirements, limitations and the next sensible step in a technical initial consultation.