North Mini Code - Cohere

2025-12-01 · Source: cohere.com via Google News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, quick

Summary

Cohere has released North Mini Code, the inaugural model in its North family of code agent models, under the Apache 2.0 license. This Mixture of Experts model features 30B total parameters with a 3B active parameter footprint, specifically trained for agentic coding. Its compact active size makes it suitable for local deployment, offering competitive performance against larger open-source models on software engineering and terminal-based agentic benchmarks. North Mini Code supports a 256K token context window and 64K max output tokens, and is designed for agentic software engineering, terminal-based agents, local/on-device coding, and general code generation. It is available for production use through Cohere's Model Vault and is free until rate limits are reached for both trial and production keys.

Key takeaway

For AI Engineers evaluating code generation models for local or agentic applications, North Mini Code presents a strong candidate. Its 3B active parameter footprint enables efficient local deployment and low-latency inference, making it ideal for on-device coding. You can integrate this Apache 2.0 licensed model into your software engineering workflows or terminal-based agents, utilizing its competitive performance on benchmarks without immediate cost, as it is free until rate limits are reached.

Key insights

A compact 3B active parameter MoE model excels in agentic coding and local deployment.

Principles

• Small active footprint enables local deployment.
• MoE architecture balances size and performance.
• Training on multiple harnesses improves generalization.

Method

The model is trained against multiple agent harnesses to ensure performance generalizes across various agent scaffolds, rather than being tuned to a single one.

In practice

• Implement repo-level code changes with SWE-Agent.
• Drive shell tools end-to-end for multi-turn tasks.
• Perform low-latency inference on local hardware.

Topics

• Agentic Coding
• Mixture of Experts
• Code Generation
• Local Deployment
• Apache 2.0 License
• Cohere Models
• Terminal Agents

Best for: AI Architect, MLOps Engineer, NLP Engineer, AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• Introducing North Mini Code: Cohere’s First Model For Developers — North Mini Code is a 30B MoE model excelling in agentic coding through cascaded SFT and RLVR.
• North Mini Code: Agentic Coding Model for Developers - Cohere — Cohere's North Mini Code is an efficient, open-source agentic coding model designed for sovereign developer control.
• How to Access and Use Qwen3-Coder-Next? — Qwen3-Coder-Next is an open-weight, agentically trained coding model optimized for local execution and low inference costs.
• Improving Bash Generation in Small Language Models with Grammar-Constrained Decoding — Constrained decoding enhances small language models' Bash command generation reliability by enforcing grammatical correctness.
• The Free Open Model Matching GPT-5.5 on Agentic Coding — Nex-N2-Pro, a free open-weights model, matches GPT-5.5's agentic coding performance on SWE-Bench Pro.
• MiniMax M2.7 Goes Open-Weight to Let You Run Agents Locally — MiniMax M2.7's open-weight release enables local deployment for complex agentic and software engineering tasks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cohere.com via Google News.