On 9 June 2026, Cohere and Cohere Labs announced North Mini Code 1.0. The first North-family release targeted software-engineering agents that need a small active footprint without giving up long trajectories.
This release brief was checked against first-party material on 10 August 2026. The date above is the public announcement date, not the date a repository was created or a third-party provider added the model. Where access or weights arrived later, that distinction is recorded below.
Release record
| Field | Verified detail |
|---|---|
| Announcement | 9 June 2026 |
| Availability or weight release | 9 June 2026 on Hugging Face and in supported coding agents |
| Release type | open-weight coding and terminal model |
| Access | Apache 2.0 BF16 weights plus official FP8 and W4A16 variants |
| Architecture | a 30B-total, 3B-active sparse MoE with 128 experts and tool-aware post-training |
| Maximum stated context | 256K input and 64K maximum output |
What changed
North Mini Code is notable for its active size and training strategy. Cohere used multiple agent scaffolds during post-training instead of optimising for one harness, then published quantisations and a model card that explains how interleaved thinking should be retained between turns. For local and private coding agents, that combination is often more useful than a larger general model with fragile tool formatting.
The practical comparison is therefore not simply whether North Mini Code 1.0 has the largest headline score. Teams need to ask whether its architecture, access terms, latency, tool behaviour and evaluation setup match the workload they actually intend to run. A model can lead one harness while losing on cost, refusal behaviour, multilingual quality or repeatability in another.
Benchmarks worth retaining
| Evaluation | Reported result | How to read it |
|---|---|---|
| Artificial Analysis Coding Index | 33.4 | Launch comparison among open models |
| SWE-Bench Pro | 40.2 in the model-card widget | Repository task result under the linked evaluator |
| SWE-Bench Verified | 67.6 in the model-card widget | Different benchmark and harness from Pro |
These are release-time results, not independently reproduced guarantees. Cohere’s headline coding index comparison and the Hugging Face evaluation widgets use different harnesses; the card also requires recent inference software and Cohere-specific parsers. Scores should remain attached to the disclosed effort setting, agent harness, tool access, timeout, context-management policy and judge model. Moving a number into a procurement sheet without those conditions creates false comparability.
Architecture and access
North Mini Code 1.0 is described as a 30B-total, 3B-active sparse MoE with 128 experts and tool-aware post-training with 256K input and 64K maximum output of stated context. Its access position at verification time is Apache 2.0 BF16 weights plus official FP8 and W4A16 variants. That wording matters: open weights, source-available weights, an API, a product preview and a research demonstration give adopters very different rights and different levels of reproducibility.
Before deployment, record the exact model identifier or checkpoint, inference stack, quantisation, reasoning setting, region, price schedule and supplier terms. If the release uses a custom licence, read the licence itself rather than relying on the word “open” in launch copy. If it is API-only, preserve the dated documentation and change-notice route because the served snapshot can change without a downloadable artefact.
What an evaluation should test next
For North Mini Code 1.0, a credible internal gate should include:
- a frozen set of representative tasks with pass, fail and abstain criteria;
- a matched baseline using the same tools, timeout, prompt budget and reviewer rubric;
- repeated runs to expose variance rather than reporting a single best attempt;
- latency, token use and total task cost alongside task success;
- adversarial, multilingual and long-context cases relevant to the real deployment; and
- rollback evidence showing the previous model can be restored safely.
The wider model change-control guide explains how to keep model, prompt, tool and corpus changes reconstructable. The AI dependency inventory guide covers the release and supplier records needed after deployment.
AIEngine verdict
North Mini Code is a strong small-active-parameter open coder. Its production value depends on preserving reasoning and tool-call history correctly, so integration tests should be treated as part of model evaluation.
This is a launch assessment, not a certification. Benchmark leadership is useful evidence of where to test; it is not authorization to place the model in a high-impact workflow without domain evaluation, security review and an accountable owner.
Primary sources
Image provenance
Hero image: Cohere Labs official release artwork via Hugging Face. The locally served WebP is a crop of the first-party release or model-card asset recorded in the repository provenance manifest.
