Leaderboard / Cohere
North Mini Code
Cohere · North · released 2026-06-17 · data checked 2026-11-01 Needs verification
North Mini Code is a 30B-parameter Mixture-of-Experts model (3B active per token) from Cohere (Other), released 2026-06-17 under the Unverified license. It doesn't have enough reported benchmarks for an index score yet. It needs about 21.4 GB of memory at 4-bit with an 8K-token context.
Cohere's first open agentic coding model. Check the license on the model card before commercial use.
Benchmarks
Each bar shows this model's score. The tick marks the best medium model on that benchmark.
No benchmark results for this release yet. Know of one? Send it.
Specifications
- Lab
- Cohere (CA · Other)
- Architecture
- Mixture-of-Experts
- Total parameters
- 30B
- Active parameters
- 3B
- Size class
- Medium
- Context window
- 256K tokens
- Input modalities
- text
- Reasoning
- Instruct (no thinking)
- License
- Unverified
- Openness
- Open weights
- Memory to run
- ~21.4 GB at 4-bit, 8K context (more sizes)
Compare with
- North Mini Code vs Granite 4.2 30B (30B, index 67.0)
- North Mini Code vs Muse Glimmer 30B (29.6B, index 72.6)
- North Mini Code vs Qwen3 30B-A3B Thinking 2507 (30.5B, index 59.3)
Sources
Something wrong here? Report it.
Run it locally
No quantized build from a well-known publisher yet; check the model card for instructions.
Memory needed: weights, context cache and runtime buffers (context cache estimated: attention layout not published).
| Quantization | 4K context | 32K context | 256K context |
|---|---|---|---|
| 16-bit F16 | 62 GB | 69 GB | 126 GB |
| 8-bit Q8_0 | 34 GB | 41 GB | 98 GB |
| 4-bit Q4_K_M | 20 GB | 28 GB | 84 GB |