Ollama setup: local and cloud model walkthrough

The presenter installs Ollama through a terminal on macOS and points out download options for Windows and Linux. He opens a project folder, explores Ollama's terminal launcher and explains that the coding assistant needs a separate installation before launch.

The walkthrough distinguishes downloaded models from cloud models that require an Ollama account. It shows where to check model sizes and discusses choosing a model that fits the computer's resources. A displayed 23 GB listing is an example from the tutorial, not a measured RAM requirement or a benchmark.

For cloud inference, the presenter reviews account plans and usage credits, signs in and connects the terminal session to his account. He approves access to the project folder and sends a greeting through the coding assistant. Cloud processing is separate from running a downloaded model on the computer; the video's credit discussion does not establish unlimited access.

The presenter then switches to a downloaded 3 billion parameter model and asks for Python code that adds two numbers. The response reads two inputs, adds them and prints the result. This demonstrates a basic generation request through the connection. The video does not show execution of that script or establish performance on larger coding tasks.