How to USE Claude Code for FREE with Ollama ( Local AI FULL Tutorial)

Getting Started with Cloud Code on Linux

Introduction to Cloud Code Installation

• The tutorial focuses on installing cloud code on Linux, applicable for various operating systems including Windows 11 and macOS.

Hardware Requirements for AI Performance

• The presenter discusses their HP gaming laptop's specs, emphasizing the need for at least 16 GB of RAM and an Nvidia GPU with 4GB of VRAM to run models like Quen 3.5 and GLM 4.7 Flash effectively.

Installing Ola

• Instructions are provided to install Ola, which is necessary for running Cloud Code. Users should visit the official website, copy the installation command, and paste it into their terminal.

• For Mac users, a DMG file is available; Windows users can follow similar steps in PowerShell. Successful installation confirms detection of the Nvidia GPU.

Verifying Ola Installation

• Users are instructed to check if Olama is running by entering a specific command and confirming that it operates on a local port through a browser URL.

Installing Cloud Code

Steps to Install Cloud Code

• The process varies by OS: simple command line for Linux/Mac or relevant PowerShell commands for Windows.

• After installation, users must add cloud code to system environment variables and verify successful installation by checking the version displayed in the terminal.

Running Local Models with Cloud Code

• To utilize local hardware power, users need to download an efficient model from Olama’s library; Quen 3.5 is recommended for low-end systems.

Creating Custom Models

Downloading Models

• A command is provided to download a model with six billion parameters that supports large context windows.

Testing Model Performance

• Users can test model performance in ripple mode; it runs decently using CPU and partial GPU acceleration while maintaining stable performance.

Configuring Context Window Size

Increasing Context Window Size

• The default context window size is limited; instructions are given to create a custom model file with an increased context size of 64K.

Verifying Custom Model Creation

• After creating the new model, users can confirm its functionality by checking if it loads successfully with the larger context window.

Connecting Model to Cloud Code

Setting Up Workspace

• Users should create a workspace folder and launch cloud code using their custom model name.

Testing Functionality

• A simple prompt test shows that responses may be slower when running locally but function correctly as expected.

Practical Applications of Cloud Code

Generating Programs

• Demonstrations include generating C programs and shell scripts successfully using tool calling features within cloud code.

Building Static Websites Using VS Code

Creating HTML/CSS Projects

• Instructions guide users through creating a new workspace in VS Code where they can launch cloud code using Quen 3.5.

Final Thoughts on Performance

• While lower-end systems may experience slower response times during generation tasks, higher-end machines will yield faster results; alternatives exist via Olama's cloud models for those unable to run locally due to hardware limitations.

How to Use Kimik 2.5 Cloud Edition for Fast Application Development

Getting Started with Kimik 2.5 Cloud Edition

• The video introduces the use of Kimik 2.5 cloud edition, which is accessible for free and operates effectively in various scenarios.

• Users are guided to run a simple command that prompts them to open a web browser and sign into their Alama account using either email or Google account.

• After logging in, users need to connect their device before returning to the terminal for further actions.

Testing Performance and Features

• The presenter demonstrates the speed of responses when using Kimik 2.5 for agent-based tasks, highlighting its efficiency.

• Users can exit the interface at any time through an exit option provided within the application.

Creating a To-Do List Application

• A practical example is shown where Kimik 2.5 is utilized in VS Code to create a simple to-do list application, completing tasks remarkably fast (30 to 40 seconds).

• The final output of the application is visually appealing, showcasing the capabilities of cloud code execution with this model.

Conclusion and Engagement

• The presenter invites viewers to share their thoughts in the comments section and encourages likes and subscriptions for more content.

• The video concludes with gratitude from KSK Royal, indicating future interactions with viewers.