North-Mini-Code-1.0

Model Summary

North Mini Code is an open weights research release of a 30B-A3B parameter model optimized for code generation, agentic software engineering, and terminal tasks.

Developed by: Cohere and Cohere Labs

• Point of Contact: Cohere Labs
• License: Apache 2.0
• Model: North Mini Code
• Model Size: 30B total; 3B active
• Context length: 256K & 64K max output

For more details about this model, please check out our blog post.

Try North Mini Code

You can try out North Mini Code before downloading the weights in OpenCode and our hosted Hugging Face Space.

Evaluation

• We used SWE-Bench Verified, SWE-Bench Pro, Terminal-Bench v2, and Terminal-Bench Hard to benchmark North Mini Code's agentic coding capabilities. For evaluation harnesses, we used the Swe-Agent harness v1.1.0 for SWE-Bench, and a simple ReAct harness employing a single terminal-use tool based on Harbor's Tmux session implementation for Terminal-Bench v2. For Terminal Bench Hard, we directly used Terminus-2, following the same methodology as the Artificial Analysis Intelligence Index to compare North-Mini-Code-1.0 with the other models. Additionally, we used SciCode and LiveCodeBench v6 as complex code-generation benchmarks outside of tool use.
• We run each benchmark with 3 different seeds and report the average benchmark performance, using temperature=1.0 and top_p=0.95. We used publicly reported scores for competitor models, either from original reports or the Artificial Analysis Intelligence Index, where available. Additionally, Gemma4’s scores for agentic coding tasks were reported by Qwen team. For benchmark results that any public report is missing, denoted by (*) in the figure, we run them internally using the recommended model configuration.

Usage

Please install transformers from the source repository that includes the necessary changes for this model. We recommend using the following set of sampling parameters for generation: `temperature=1.0`, `top_p=0.95`.

# pip install transformers
from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "CohereLabs/North-Mini-Code-1.0"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

prompt = "Write a python program to check if a string is a palindrome or not."

# Format message with the North-Mini-Code-1.0 chat template
messages = [{"role": "user", "content": prompt}]
input_ids = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt",
)

gen_tokens = model.generate(
    **input_ids,
    max_new_tokens=1024,
    do_sample=True,
    temperature=1.0,
    top_p=0.95
)

gen_text = tokenizer.decode(gen_tokens[0])
print(gen_text)

You can also use the model directly using transformers pipeline abstraction:

from transformers import pipeline
import torch

model_id = "CohereLabs/North-Mini-Code-1.0"

prompt = """Given a list of unique words each of size k and an n sized word, w, where n is a multiple of k,
Write a program in python to determine the number of unique combinations of words in the list that can be concatenated to form an anagram of the word w.
"""

pipe = pipeline(
    "text-generation",
    model=model_id,
    torch_dtype="auto",
    device_map="auto",
)

messages = [
    {"role": "user", "content": f"{prompt}"},
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)

outputs = pipe(
    messages,
    max_new_tokens=1024,
    do_sample=True,
    temperature=1.0,
    top_p=0.95

)

print(outputs[0]["generated_text"][-1])

Model Details

Input: Text only.

Output: Model generates text.

Model Architecture: North-Mini-Code-1.0 is a decoder-only Transformer-based sparse Mixture-of-Experts model. It uses an efficient attention implementation, interleaved between sliding-window attention with RoPE and global attention with no positional embeddings, in a 3:1 ratio. The feed-forward block is an MoE block with 128 experts, of which 8 are activated per token. Each expert block is an FFN block with SwiGLU activation. The router applies a sigmoid activation function to the logits before the top-k selection. We also use a single dense layer before the sparse layers. North-Mini-Code-1.0 was post-trained using a two-stage cascaded supervised fine-tuning (SFT) followed by reinforcement learning with verifiable rewards (RLVR), focusing on agentic coding. For more technical details, please check out our blog post.

Context Length: North-Mini-Code-1.0 supports a context length of 256K & 64K output length.

Tool Use Capabilities:

North-Mini-Code-1.0 has been specifically trained with tool-use capabilities for agentic coding.

Tool use with North-Mini-Code-1.0 is supported through chat templates in Transformers. We recommend providing tool descriptions using JSON schema.

Tool Use Example [CLICK TO EXPAND]

# Define tools
tools = [{
  "type": "function",
  "function": {
    "name": "bash",
    "description": "Execute a bash command in the terminal.",
    "parameters": {
      "type": "object",
      "properties": {
        "command": {
          "description": "The bash command to execute.",
          "type": "string"
        }
      },
      "required": ["command"]
    },
  }
}]

# Define conversation input
conversation = [{"role": "user", "content": "Find out if there is any json file in this folder"}]

# Get the Tool Use prompt
input_prompt = tokenizer.apply_chat