The model scored 33.4 on the Artificial Analysis Coding Index.
We made a small coding model. Its open source apache 2.0. Now more than ever i think this tech needs to be built in public so that those using it are in control. Try it out if you want a small and efficient coding model.
Many users praised Cohere's open-sourcing of the North Mini Code model for adding useful production-ready tools to the ecosystem and thanked the team, while a few criticized flaws shown in the launch video.
Introducing Cohere's first open-source coding model: North Mini Code
Small & efficient, designed for agentic performance and built for community input.
this model is the opposite of mythos.
Its small, cost effective, apache 2.0, and locally deployable. This is the way LLMs should go.
small, open source, transparent and sovereign vs large, expensive, proprietary and hegemonic
Introducing Cohere's first open-source coding model: North Mini Code
Small & efficient, designed for agentic performance and built for community input.
We encourage developers to share their builds with us and give feedback to shape future iterations. Let’s shape the future of sovereign AI together.
Download: https://huggingface.co/CohereLabs/North-Mini-Code-1.0
Cohere just released North Mini Code, a small 30B parameter (3B active) open weights coding model that scores 27.6 on the Artificial Analysis Intelligence Index
Less than a month since @cohere's last model release, Command A+, has launched another open weights model that is optimized for coding, and much smaller at 30B total parameters and 3B active parameters.
Key Takeaways:
➤ Achieves 27.6 on the Artificial Analysis Intelligence Index, above gpt-oss-20B (high) at 24.5 and just below Mistral Small 4 (119B parameters, 6.5B active) at 27.8
➤ Scores competitively on the Artificial Analysis Coding Index (weighted average of Terminal-Bench Hard and SciCode) against open weights models in its size class, scoring 33.4, significantly above GLM-4.7-Flash at 25.9, and below Qwen3.6 35B A3B at 35.2. However, it underperforms on non-coding agentic tasks, scoring 14% on GDPval-AA and 37% on 𝜏²-Bench Telecom
➤ On Cohere’s API, North Mini Code is faster than several comparable open weights models of its intelligence and size class (~199 output tokens per second)
➤ North Mini Code is a text-only 30B total parameter and 3B active parameter model, and is open-sourced under the Apache 2.0 license
We open-sourced a feisty small agentic coding model.
- 30B total, 3B active - 256K total context - Compatible with @opencode - Apache 2.0. Weights on @huggingface
Introducing Cohere's first open-source coding model: North Mini Code
Small & efficient, designed for agentic performance and built for community input.
Small: 30 billion parameters, 3B active.
Efficient: Benchmarks to 33.4 on the Artificial Analysis Coding Index, competitive among similar sized models.
Open Source: Apache 2.0 license so developers can experiment, test, and build their way.
Learn more: https://cohere.com/blog/north-mini-code
codehere
Introducing Cohere's first open-source coding model: North Mini Code
Small & efficient, designed for agentic performance and built for community input.
@nickfrosst More info:
@cohere Nice! Great to see another open source model released. 🙌
North Mini Code scores 14% on GDPval-AA and 37% on 𝜏²-Bench Telecom, resulting in an overall weighted score of 21.7 on the Artificial Analysis Agentic Index
One of many more great open source models by Cohere. Optimized for local coding agents.
Cohere is on a mission to enable sovereign AI, and being able to locally host models is part of this mission.
Introducing Cohere's first open-source coding model: North Mini Code
Small & efficient, designed for agentic performance and built for community input.
North Mini Code uses more output tokens to complete the Artificial Analysis Intelligence Index evaluations suite than most comparable models of its size and intelligence
Full intelligence evaluations breakdown below:
In our pre-release speed testing, North Mini Code performed above several comparable open weights models of its intelligence and size class (~199 output tokens per second)
@cohere Wow! This is another cool release!
See Artificial Analysis for further details and benchmarks: https://artificialanalysis.ai/models/north-mini-code
@cohere congrats! https://huggingface.co/blog/CohereLabs/introducing-north-mini-code
@nickfrosst Props. Will try it out in the coming days. I missed what harness are you using?
@ArtificialAnlys @cohere Qwen3.6-35B-A3B is such a beast. The loss compared to 27B on that benchmark is so small that it basically doesn't exist when you consider the 35B-A3B is 2x or more faster than the 27B.
@cohere You dont need an h100 to run this 🤔
The model scored 33.4 on the Artificial Analysis Coding Index.
We made a small coding model. Its open source apache 2.0. Now more than ever i think this tech needs to be built in public so that those using it are in control. Try it out if you want a small and efficient coding model.
