Curso de OLLAMA 2026: Tu propia Inteligencia Artificial en LOCAL (Instalación, Uso y Casos Reales)

This section introduces Oyama as a tool for running powerful language models locally, eliminating the need for cloud services or external APIs. It highlights Oyama's ability to run models without a dedicated GPU, relying on CPU and RAM, and discusses how PC specifications affect model choice. The video outlines the topics to be covered: an introduction to Oyama, installation, model installation, practical examples, and setting up a graphical interface.

This part explains that Oyama simplifies the execution of large language models (LLMs) on Windows, Mac, and Linux by managing downloads, versions, and execution. It clarifies that Oyama is the software that manages LLMs, similar to Docker managing containers. The importance of model parameters, which determine their 'intelligence' and size, is discussed. Practical advice is given on selecting models based on RAM requirements (8GB for 7B models, 16GB for 13B models, 32GB for 33B+ models), and how to browse available models and their capabilities on Oyama's website.

The installation process for Oyama is demonstrated. Users can download the appropriate installer for Mac, Linux, or Windows from the Oyama website. The video mainly focuses on the Windows installation as an executable wizard, showing how Oyama integrates as a background service or starts with the operating system.

This section details how to download and manage models using command-line commands. The `ollama pull` command is used to download models from a central repository. A lightweight model, 'llama 3.2 1B', is chosen for demonstration purposes to accommodate systems with limited resources. The `ollama list` command is shown to verify installed models, and `ollama show` to get detailed information about a specific model.

The video demonstrates how to run an installed model using `ollama run` for an interactive chat session. Examples are provided for various use cases: acting as a technical assistant to answer Linux command queries, explaining complex commands, and summarizing text for junior developers. It also illustrates how the model can filter messages for support departments. The limitations of simpler models, such as their tendency to over-explain or perform better in English, are noted.

This part focuses on using a more powerful model for code generation. The 'Qwen Coder' model, optimized for code development, is chosen. The process of downloading this larger model (using `ollama pull`) is shown, followed by running it. A detailed prompt is used to ask the AI to generate a simple web application for a medical clinic using HTML, CSS, JavaScript, and a Node.js Express backend. The generated code is then copied into Visual Studio Code, and the AI is prompted to provide instructions on how to install and run the application locally.

The video introduces graphical user interfaces for interacting with Oyama. First, it shows the built-in Windows GUI for Oyama, demonstrating how to select models, send messages, and upload files for analysis. Then, it introduces Open Web UI, an extensible self-hosted user interface designed to provide a ChatGPT-like experience locally. It highlights Open Web UI's features, including multi-model support, document processing (RAG), user management, and external tool integration. The section prepares for the installation of Open Web UI.

This segment provides a step-by-step guide on installing Open Web UI using Python, emphasizing the creation of a virtual environment to manage dependencies. The `pip install open-webui` command is executed within the virtual environment. After installation, `open-webui serve` is used to launch the server, accessible via `localhost:8080`. The video demonstrates logging into the Open Web UI, selecting an Oyama model, and interacting with it through a chat interface. It concludes by encouraging users to explore Open Web UI's features and noting that similar tools like Onix and MSTY also exist.