North Mini Code, Explained Simply
What North Mini Code is, what it does well, and what it costs โ released 2026-06-09 by Cohere.
TLDR
- Cohere just released its first AI model specifically built to help with computer programming.
- It is designed to act like a smart agent that can help you write and fix code.
- It is small enough that you can run it on your own computer, rather than relying on a giant server.
- It is completely free to download and use.
Want a heads-up when Cohere updates?
We email you the moment a new Cohere version drops โ plus plain-English release notes like these. No spam.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Cohere's North Mini Code is an AI designed specifically to help people write, fix, and manage computer code. It is built for developers and students who want a smart programming assistant that can actually run on their own local computer hardware instead of requiring a massive, expensive server.
What it does well
- Agentic coding: Instead of just giving you a snippet of text, it acts like an "agent" that can help carry out multi-step coding tasks to solve problems.
- Runs locally: It uses a "Mixture of Experts" design. This means even though the whole AI is made of 30 billion parts (parameters), it only uses 3 billion at a time. This keeps it lightweight enough to run on a good personal computer.
What it costs and what it can handle
- Price: It is totally free to download because it uses an Apache 2.0 license, which means anyone can use or modify it.
- Context Length: It can read up to 256,000 "tokens" at once (which is roughly 200,000 words of code or text) and can generate up to 64,000 tokens in its reply. This is a huge amount, letting you feed it multiple large files of code at the same time.
Worth knowing
- Where to get it: You can use it through Cohere's Chat V2 API (a way for apps to talk to the AI) or download the open weights from Hugging Face (a popular website for sharing AI models).
- For big projects: If a company wants to use this for heavy, real-world production, they will need to use something called a "Model Vault" to deploy it safely.
This is our plain-language summary. Read the complete, official notes on the Cohere official changelog โ.
