North Mini Code: benchmark results
Cohere Labs' 30B-A3B mixture-of-experts research release tuned for code generation and agentic software-engineering and terminal tasks, with up to 256K context. Provider: Cohere. Released 2026-06-09. Access: Open.
Aggregate Rating 1579 ± 22, rank #519 of 1656 rated models, from 86 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Omniscience - Software Engineering (SWE) - Rust | 56 | Accuracy (%) | 74.1 |
| AA Terminal-Bench Hard | 31.06 | Accuracy (%) | 73.7 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 24 | Accuracy (%) | 72.5 |
| AA Omniscience - Software Engineering (SWE) - JavaScript | 34.55 | Accuracy (%) | 72 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 26.67 | Accuracy (%) | 69.3 |
| AA IFBench | 57.55 | Accuracy (%) | 69 |
| AA Omniscience - Software Engineering (SWE) - Go | 22 | Accuracy (%) | 68.4 |
| AA Omniscience - Software Engineering (SWE) - Java | 18 | Accuracy (%) | 66.9 |
| AA Omniscience - Software Engineering (SWE) - HTML | 34 | Accuracy (%) | 65.9 |
| AA Omniscience - Software Engineering (SWE) - Python | 21 | Accuracy (%) | 60.9 |
| AA Omniscience - Software Engineering (SWE) - Swift | 36 | Accuracy (%) | 60.6 |
| AI Chess Leaderboard (Reasoning) | 756 | Elo | 58.5 |
Interactive version: theaggregate.ai/model?slug=north-mini-code · How It Works · Data refreshed daily, snapshot 2026-10-20.