Cohere · Apache 2.0
North Mini Code
Cohere’s 30B-A3B research model for code and terminal tasks.
How much memory it needs
The model file, plus memory for the conversation, plus about 1 GB for the app. Longer conversations need more.
| Size | Version | File | At 8K | At 32K | At 128K |
|---|---|---|---|---|---|
| 31B (3B active) | Q4_K_M | 18.7 GB | 20.0 GB | 20.9 GB | 24.4 GB |
| Q8_0 | 32.4 GB | 33.7 GB | 34.6 GB | 38.1 GB |
Which computers run it
The best size for each machine at an 8K-token conversation. Speeds are estimates from memory bandwidth.
| Computer | Best version | Fit | Speed |
|---|---|---|---|
| Laptop or PC, no graphics card, 32 GB | 31B Q4_K_M | Runs well | 15–30tok/sSmooth |
| RTX 4060 · 8 GB | 31B Q4_K_M | Spills into system RAM | 19–35tok/sSmooth |
| RTX 3060 · 12 GB | 31B Q4_K_M | Spills into system RAM | 25–50tok/sSmooth |
| RTX 5060 Ti · 16 GB | 31B Q4_K_M | Spills into system RAM | 35–70tok/sFast |
| RTX 4090 · 24 GB | 31B Q4_K_M | Runs well | 110–210tok/sFast |
| RTX 5090 · 32 GB | 31B Q4_K_M | Runs great | 140–280tok/sFast |
| Mac with M4 · 16 GB | 31B Q4_K_M | Too big | — |
| Mac with M4 Pro · 48 GB | 31B Q8_0 | Runs well | 30–65tok/sFast |
| Mac with M5 Max · 128 GB | 31B Q8_0 | Runs great | 50–110tok/sFast |
| Ryzen AI Max+ 395 · 128 GB | 31B Q8_0 | Runs great | 25–50tok/sSmooth |
| DGX Spark · 128 GB | 31B Q8_0 | Runs great | 25–55tok/sFast |
| RTX PRO 6000 · 96 GB | 31B Q8_0 | Runs great | 110–220tok/sFast |
| Mac Studio with M5 Ultra · 256 GB | 31B Q8_0 | Runs great | 50–100tok/sFast |
| Mac Studio with M3 Ultra · 512 GB | 31B Q8_0 | Runs great | 40–85tok/sFast |
What it is good and bad at
Good at
- Code generation and terminal tasks
- Fast: about 3B parameters work per word
Not good at
- Research release with low adoption so far
Start it
Install Ollama, then run one of these in a terminal. LM Studio users can search the same name in the app.
ollama run north-mini-code-1.0:latest