Video Summary
Install Node.js and OmniRoute, then open the OmniRoute local UI to add providers.
Connect free providers (OpenRouter, Kiro, Antigravity, Nvidia NIM) and import only free models.
Create a free model combo in Claude Code, prefer round-robin strategy for automatic failover.
Generate an OmniRoute API key, configure Claude Code environment variables, and connect to the combo.
Test models in OmniRoute, run the Claude Code project (npm run dev) and verify the generated website.
OmniRoute aggregates many AI providers and exposes their free models as endpoints. You import free models (e.g., via OpenRouter, Kiro, Antigravity, Nvidia NIM), create a model combo, generate an OmniRoute API key, and point Claude Code at that combo so requests are routed across free providers.
You need Node.js installed and an editor/terminal (the video uses VS Code). Also prepare free API keys/accounts for providers you plan to connect (OpenRouter, Kiro, etc.).
When building a combo in OmniRoute/Claude Code, choose a round-robin (or failover) strategy and add multiple free models. OmniRoute will automatically route to the next working model if one hits quota or fails.
Test individual models in the OmniRoute UI, create and test a combo, copy the generated OmniRoute API key into Claude Code environment variables, run the app, and use commands like npm run dev to confirm the generated website functions.
"Cloud code is very expensive. If you have used it, you already know this."
Many users are familiar with the high costs associated with cloud code services, especially when reaching usage limits just when they need the resource the most.
Anthropic only offers a solution by requiring additional payments when these limits are reached.
"What if I told you that you could get 1.6 billion free tokens on Cloud code every single day?"
The video introduces a way for users to acquire 1.6 billion free tokens daily for cloud code usage without incurring any costs.
The creator shares a personal project created using this setup to illustrate its effectiveness and encourages viewers to follow along for setup instructions.
"So, let's start with installing Omni Route first."
The setup process begins with installing a tool called Omni Route, which is explained step by step.
Users need to visit the omniroute.online website, where they can start the installation by simply copying a command and pasting it into the terminal.
"To run this command, you need to install Node.js first."
The installation of Omni Route requires an additional software called Node.js, which must be installed prior to using the main command.
Users should follow specific steps to download and install Node.js on their systems, ensuring compatibility with their operating system.
"We need to connect some providers to build any project."
Once Omni Route is running, the next step is to connect various free AI providers to access powerful AI models.
The total number of available providers is highlighted, with 126 of them being accessible for free, allowing users to enhance their projects significantly.
"The number one that we are going to set up is OpenRouter."
OpenRouter is identified as the primary provider to connect, and users can easily obtain an API key by signing up for a free account on its official website.
Detailed steps on how to create an API key are provided, including naming the key and selecting an expiration time.
"Test if the models are working or not."
After importing the models using the API key, users are encouraged to test them to ensure they are functional.
A few models may not work, but Omni Route will only utilize those that are compatible, streamlining the workflow for users.
"The first provider we need to use is called Kiro AI."
The video continues by guiding users through the process of connecting to various other AI providers, including Kiro AI and Anti-gravity, which offer high-quality models.
Each provider's connection steps are explained, ensuring that users have access to a broad range of powerful AI tools.
"Now, let's install Claude code onto our system."
Claude Code is introduced as another essential tool in the setup, and instructions are provided for its installation process, dependent on the user's operating system.
The creator emphasizes troubleshooting tips for confirming a successful installation, particularly regarding environmental variables that might affect recognition in the terminal.
"Whenever you are using Claude code, you need to launch OmniOut first."
To ensure the system runs smoothly, users are reminded to launch OmniOut after every restart, enabling the API keys to function correctly.
A note on the local operation of the system versus the cloud models is mentioned, emphasizing the importance of maintaining a proper setup for optimal performance.
"Click on free stack to choose models that are completely free and name your combo whatever you want."
To start, navigate to the combo section of the platform and click on "free stack" to select models that incur no charges. You can assign any name to your combo; for this example, "Amedia-AI" will be used for future configuration in Cloud Code.
After naming the combo, proceed to select your provider, with "Antigravity" being chosen for this instance. You will be presented with a list of available free models.
The arrows next to the models indicate their loading priority—alter this to set your preferred order. For instance, if you wish to prioritize "Cloud Sonnet 4.5" over "Flashlow," simply adjust their positions in the list.
"I suggest you choose the round robin strategy for automatically rerouting to the next best model."
You also have the option to add models not present in your initial provider by selecting "choose model" and then clicking on "add step." For instance, if you want to include "Opus 4.6" and "Sonnet 4.6," select each model and add them accordingly.
After setting up your preferred models, move on to the strategy step. It will default to "priority," but switching to "round robin" is recommended for better performance, as it automatically selects the next optimal model when needed. Once completed, click on "next" to proceed.
Click on "create combo" to finalize your setup, and then test the combo to ensure selected models are functioning correctly.
"Copy this API key as it won't be shown again, so store it somewhere safe."
Once the combo is created, the next step involves generating an API key. Name the key anything you wish—"AI" can be used for simplicity. Click "create API key," and be sure to copy the key since it is crucial for future operations. Store it securely, as it won't be retrievable later.
You will then transition back to working with Cloud Code. Open your IDE (like VS Code) and set up a new terminal session to input commands with necessary modifications outlined for successful implementation.
"As this model stops working or the quota is full, it will switch to the next model automatically in milliseconds."
In the terminal, you'll enter specific commands to reference the newly created combo. Ensure you replace necessary segments with your combo's name and your Omni route key accurately. Hitting enter will initiate the setup.
Once prepared, the terminal will prompt for additional setup questions, where you can opt for recommended settings. Verify that your combo, "Amedia AI," is operational by testing various inquiries.
Monitor your usage through the control center interface—success rates and quotas are displayed here. Should some models fail, the system swiftly transitions to alternate models without noticeable delay, ensuring efficiency.
"Cloud has created this folder, and it has created the website inside this."
Test the capabilities of your setup by asking it to generate applications or websites. Input your extensive prompt to challenge the system. When you observe the results, note that a website will be created within a designated folder.
Execute "npm run dev" within that folder to launch the website and assess its functionality and design. The website should meet your requirements, and adjustments can be made based on your preferences.
