North Mini Code First Results

Recently Cohere AI released an open-weights coding model called North Mini Code. This Mixture-of-Experts coding model has 30 billion parameters and 3 billion active parameters. I downloaded a 4-bit quantization by Unsloth with file name North-Mini-Code-1.0-UD-IQ4_NL.gguf; it was 15.5 GB.

North Mini Code

After compiling llama-cpp, I was able to run North Mini Code on llama-server on my first attempt. All went well. I opened OpenCode and began trying to use the LLM to make some small refactoring changes to my Rust program.

I found that:

• North Mini Code is a truly separate model, one trained from scratch, and also trained for OpenCode use.

• It is from Canada.

• The model was able to make the required changes to my Rust code base, but it did not seem superior to Qwen 3.6 35B, and it required a lot of repeated prompting.

Overall my experience was positive and if Qwen 3.6 35B was unavailable for some reason I would be more likely to use North Mini Code. It seemed able to do tool calls and code edits. The changes I had it make to my code compiled correctly and everything was fine. I feel like it summarizes things too much, so I might like using it more if I had a much faster GPU.