Build ANYTHING for FREE with OpenAI’s New OSS Models & n8n (No-Code Tutorial)

Key Concepts

• GPTO OSS models (Open Source): OpenAI's open-source GPT models, including a 120 billion parameter model and a 20 billion parameter model.
• Ollama: A tool for running open-source large language models locally.
• Naden: An AI automation platform.
• Open Router: A platform for accessing various AI models through a single API.
• Docker Desktop: A containerization platform for running applications in isolated environments.
• Virtual Private Server (VPS): A virtualized server that provides dedicated resources.
• AI Agent: An automated system designed to perform specific tasks using AI.
• AI Workshop Community: A community focused on learning how to monetize AI and build AI automation solutions for businesses.

GPTO OSS Models Overview

OpenAI has released two new open-source models, the GPTO OSS models, making GPT technology accessible for free and locally hostable. The models include a 120 billion parameter model and a 20 billion parameter model. The 20 billion parameter model is suitable for most users, especially those with Macs or desktops, and can be run locally.

• 120 Billion Parameter Model: A large model that requires significant computational resources (GPU) to run effectively.
• 20 Billion Parameter Model: A more accessible model that can be run on local machines with moderate resources. It is comparable to OpenAI's GPT-3 and GPT-4 Mini in terms of reasoning, knowledge, competition, and math capabilities.

Method 1: Using Open Router on a Virtual Private Server (VPS)

This method involves hosting Naden on a VPS (e.g., Hostinger, Digital Ocean) and connecting to the GPTO OSS model through Open Router.

Step-by-Step Process:

1. Install Naden on a VPS:
   • The video recommends Hostinger's KVM2 plan as an affordable and easy option. A link to Hostinger is provided in the description.
   • Select a hosting period (24 months is suggested for the best deal).
   • In the application tab, choose Naden.
   • Use the coupon code "AI workshop" for an additional 10% discount.
   • Complete the registration and payment process.
   • Access the Naden dashboard after installation.
2. Connect Naden to Open Router:
   • In Naden, create a workflow and add an AI agent with a chat message trigger.
   • Select Open Router as the chat model provider.
   • Create a new credential in Naden using the Open Router API key.
3. Obtain Open Router API Key:
   • Create an account on Open Router.
   • Navigate to the "Keys" section and generate a new API key.
   • Copy the API key and paste it into the Naden credential setup.
4. Select the GPTO OSS Model in Naden:
   • In the chat model settings, choose the GPTO OSS 20 billion parameter model (free version).
5. Test the Connection:
   • Send a test message through the Naden chat interface to verify the connection to Open Router and the GPTO OSS model.
   • Check the Open Router activity log to confirm the request was processed.

Benefits:

• Easy to set up and use.
• Private, as the model is hosted on a virtual machine.
• Offloads computational burden from the local machine.

Method 2: Using Docker Desktop

This method involves running the GPTO OSS model locally using Docker Desktop and Olama.

Prerequisites:

• Docker Desktop installed and running.
• Naden self-hosted AI starter kit installed (link to a step-by-step tutorial provided in the video description).

Step-by-Step Process:

1. Verify Docker Setup:
   • Ensure that the Naden containers (Naden, PostgreSQL, Quadrant, and Olama) are running in Docker Desktop.
2. Pull the GPTO OSS Model using Olama:
   • Open the Docker Desktop app and access the terminal for the Olama container.
   • Use the command olama pull gpto-oss:<model_size> (e.g., olama pull gpto-oss:20b) to download the desired model. The model name can be found on the Olama website (ola.com).
3. Connect Naden to Olama:
   • In Naden, create a new credential for Olama. The local host address (11434) should be automatically populated based on the Docker setup.
4. Select the GPTO OSS Model in Naden:
   • In the chat model settings, the newly pulled GPTO OSS model should now be available for selection.

Benefits:

• Full control over the model and data.
• No reliance on external services like Open Router.

Drawbacks:

• Requires a more powerful local machine with sufficient unified memory.
• Can slow down the computer if the model is too large.

AI Workshop Community and Monetization

The video promotes the AI Workshop community, which offers resources and training on how to monetize AI skills and build AI automation solutions for businesses.

• Five-Week Accountability Program: A program designed to help individuals start their own AI agency and become AI consultants.
• AI Agency Experience: The AI Workshop shares its own experiences in running an AI agency, including pricing strategies, client acquisition, and discovery call techniques.
• Voice AI Certification Course: A partnership with retail to offer a certification course on voice AI.

Conclusion

The video provides a practical guide to using OpenAI's GPTO OSS models, offering two distinct methods: leveraging Open Router on a VPS for ease and privacy, and running the models locally with Docker Desktop for full control. It emphasizes the accessibility of the 20 billion parameter model and highlights the potential for monetization through the AI Workshop community. The choice between the two methods depends on the user's technical expertise, available resources, and desired level of control.