01What is North Mini Code and How Does It Work?
North Mini Code is a new free agentic coding model made by Cohere. You can use it through a free API or run it locally on your own computer. It is designed to help AI agents write code, run tools, and build things for you.
It works by understanding a task you give it, like scheduling an event, and then performing the steps automatically. In this episode, Julian plugged it into Hermes Agent and asked it to schedule a Japanese practice session. North Mini Code did it quickly and correctly.
It is a small model, which makes it very fast. It uses a mixture of experts design with 30 billion total parameters, but only 3 billion are active at a time. This keeps it efficient while still being powerful.
02Key Features of North Mini Code
North Mini Code comes with several features that make it useful for AI agents.
- Free to use both on the API and locally.
- 256k context window, so you can give it a lot of information at once.
- Optimized for agentic coding, software engineering, and terminal tasks.
- Can be run locally with Ollama, no cloud needed.
- Works with tools like Hermes Agent, OpenClaw, Claude Code, CodexApp, and OpenCode.
- Apache 2.0 license, so you can keep everything you build with it.
Because it is a small model, it runs very fast. It also outperforms Gemma 4 on some benchmarks.
03How to Set Up North Mini Code with Hermes
Setting up North Mini Code is easy. In this episode, Julian showed two main ways: using the free API or running it locally.
- Get a free OpenRouter key from the OpenRouter website.
- Open your terminal.
- Type hermes model and select Open Router from the list.
- Choose North Mini Code as your model.
- Alternatively, install Ollama with a single terminal command.
- Pull the North Mini Code model using the command shown in the episode.
- Run it inside Hermes Agent with the provided terminal command.
That is all you need to do. It takes about two minutes to set up. Julian also recommends creating a separate agent profile for North Mini, so you can switch between models easily.
04How to Use North Mini Code in Practice
Once North Mini Code is set up, you can use it for many tasks. In the episode, Julian demonstrated scheduling a Japanese practice session. He gave the command to Hermes, and North Mini Code handled the rest, creating a scheduled task.
You can also use it with the Agent Operating System. This lets you select between different models, like KimiK 2.7, Quen 3.7, and North Mini. Each has its own conversation history, so you can test them side by side.
If you want to build a team of agents, you can use a Kanban board. You can set up a board just for North Mini and give it tasks. It will automatically triage the tasks and build them step by step.
Everything you build is saved in a workspace. You can preview your work and come back to it later. This makes it easy to see what your agents have done.
05Benchmarks and Comparisons Shown
In this episode, Julian showed how North Mini Code compares to other models. It is a 30 billion parameter mixture of experts model with 3 billion active. This small size makes it fast.
Julian says it outperforms Gemma 4 on some benchmarks. He also compared it to Quen 3.6 and Gemma 4, so those are the models to look at if you want to understand its performance.
The quote from Cohere in the episode says it is optimized for code generation, agentic software engineering, and terminal tasks. That makes it perfect for Hermes Agent.
Since it is free, you can run it 24/7 without worrying about running out of tokens. Julian says that is a big advantage over paid models.
06Tips, Limits, and Mistakes to Avoid
North Mini Code is free, but there are some things to keep in mind.
- It is a small model, so it may not be as powerful as premium models for complex jobs.
- Use it for small jobs to save your paid tokens for harder tasks.
- Separate agent profiles by API so you can switch easily if one goes down.
- Test models side by side with the same tasks to see which works best.
- Don't be afraid to experiment with free models because there is no risk.
Julian also warns about the old way of rationing tokens. With a free API, you can run agents all day and night. You can spin up several agents in parallel without worry.
If you are not technical, don't worry. Julian says you don't need to be a coder to build with AI agents anymore. He has 184 pages of testimonials from non-technical people setting up these systems.
07Who Should Use North Mini Code and Why
North Mini Code is for anyone who uses AI agents and wants to save money. If you run out of tokens quickly, this free model is a great backup. You can plug it into Hermes and keep working without stopping.
It is also good for people who want to run a 24-hour AI agent. Because it is free, you can let it run all night without worrying about costs. Julian says that is a big change from the old way of rationing tokens.
If you are new to AI agents, free models are the best to experiment with. You can try new things and see what happens without losing money.
For experienced users, North Mini Code lets you route small jobs to free models and save your premium models for the important work. That is a smart way to use your resources.
08How to Get More Help and Resources
If you want to learn more, Julian mentions the AI Profit Boarding community. Inside, you get access to the Agent Operating System, which includes Hermes and other models in one dashboard. You also get the setup walkthrough, four weekly coaching calls, and daily tutorials.
There is a 30-day roadmap, every prompt, and the Obsidian memory setup for all agents. You also get a zip file with the full agent operating system.
Julian says the community has 3,600 members who are building systems like this. You can ask questions and get help in real time, since Julian answers personally.
You can join by going to AIProfitBoarding.com or using the link in the description. You can also find people building with AI agents in your local area through the map feature.
