# North Mini Code 1.0 (slim)

> North Mini Code 1.0 is Cohere's open-weight coding agent model, a mixture of experts of 30B parameters with 3B active, trained for tool use, with interleaved thinking and a 256K context. Apache 2.0. Cohere's 4-bit build is 19 GB.

- Full: https://www.anchorterminal.com/tools/north-mini-code-1-0.md (~9,800 tokens) · this version ~1,580 tokens · JSON https://www.anchorterminal.com/tools/north-mini-code-1-0.json · canonical https://www.anchorterminal.com/tools/north-mini-code-1-0
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-10 · pricing https://www.anchorterminal.com/tools/north-mini-code-1-0#pricing · reviews https://www.anchorterminal.com/tools/north-mini-code-1-0#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**C · 61/100 · rank #80 of 160 open-weight models · #63 in Open-weight models · not agent-ready · confidence low**


Assessment: A coding agent model that fits one 24 GB GPU at 4 bits, trained for tool use, with a 256K context, Apache 2.0 and four released runtimes. It's a research release and reads like one: nothing on limitations, safety or data, the card's run instructions are out of date, and Cohere hasn't answered a question on the repository.

## Facts

- Kind: Open-weight model · vendor: Cohere · category: Open-weight models · legal entity: Cohere Inc. · provenance 75/100
- Packages: huggingface `CohereLabs/North-Mini-Code-1.0`, huggingface `CohereLabs/North-Mini-Code-1.0-fp8`, huggingface `CohereLabs/North-Mini-Code-1.0-w4a16`
- Auth: None · pricing: Free · x402: no · licence: Apache-2.0
- Benchmark probes: not run yet, so Performance is pending
- Surface graded: The open weights at huggingface.co/CohereLabs/North-Mini-Code-1.0 and the fp8, w4a16 and eagle repositories beside it, read from the model card, the files and their metadata, the config, the chat template, the commit history and the runtimes' own code at their latest releases on 10 November 2026. Cohere's company facts come from our other Cohere listings. The blog post and the Space weren't read. Nothing was downloaded or run
- Model: 30,484,303,872 parameters in the files; the card says 30B with 3B active. A sparse mixture of experts with 128 experts and 8 active per token, interleaving sliding-window and global attention three to one. Text in and text out. Cohere2MoeForCausalLM
- Files: 49 safetensors files in BF16 (60.97 GB), chat_template.jinja, config and tokenizer files and evaluation result files. No pickle files and no Python. Hugging Face lists a SHA-256 for the large files
- Licence: Apache 2.0 in the metadata of every repository and on the card. There is no LICENSE file
- Runtimes: The card documents Transformers from source and vLLM from its main branch with Cohere's melody library for parsing. Released vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 have the cohere2_moe model, and llama.cpp b12242 has the cohere2moe architecture
- Context: 256K tokens with up to 64K of output, per the card. config.json gives 500,000 positions, and the card's vLLM command sets a 320,000-token model length
- Tools and reasoning: The card shows tool use through the chat template, with tools described in JSON schema, the model's tool call and thinking added back to the history, and the result appended as a dictionary with the tool role. On vLLM it names the cohere_command4 tool-call and reasoning parsers. Thinking is interleaved with tool calls, and the template has a reasoning_effort setting
- Hardware: No memory figure on the card. Its vLLM command uses two GPUs. The 4-bit build is 19.4 GB
- Training data: Not described. The card gives the training method, two stages of supervised fine-tuning then reinforcement learning with verifiable rewards, and points to a blog post that wasn't read

- Scores: Capability pending, Reach 85, Risk 60, Agent interface 85, Documentation and upkeep 36 · total over the 7 assessed categories
- Why:  · Reach, Memory 25, the smallest build from Cohere is the w4a16 repository, 19.38 GB, and 19.38 GB × 1.2 = 23.3 GB, within 24 GB. · Risk, File safety 25, safetensors only, no .bin, .pt or .pth files and no Python in the repository. · Agent interface, Chat template 20, chat_template.jinja in the repository (12,397 characters). · Documentation and upkeep, Model card 11, Intended use, the architecture, the parameters and the context are on the card.
- Sources: 15, open questions: 5, both in the full twin
- Capabilities: model.open-weights
- JSON: https://www.anchorterminal.com/api/v1/tools/north-mini-code-1-0.json
- Its vendor verifies the listing with `verify_listing` or `POST https://www.anchorterminal.com/api/v1/verify` (the full twin says how); no effect on the grade.
- Support: Discord https://discord.com/invite/co-mmunity; https://cohere.com/help-center; more in the full twin

## Before you call it

1. Serve it with vLLM 0.33.0 or later and `--tool-call-parser cohere_command4 --reasoning-parser cohere_command4 --enable-auto-tool-choice`, with `cohere_melody` installed
2. Pass the model's thinking back with its tool calls on every turn. The card says results drop without it
3. Describe tools in JSON schema and append tool results as a dictionary with the tool role
4. Use temperature 1.0 and top_p 0.95
5. For llama.cpp the GGUF files are Unsloth's or bartowski's conversions, not Cohere's
6. It's text only. Don't send images

## Connect

```bash
pip install "vllm>=0.33.0" "cohere_melody>=0.9.0"
vllm serve CohereLabs/North-Mini-Code-1.0 -tp 2 --max-model-len 320000 --tool-call-parser cohere_command4 --reasoning-parser cohere_command4 --enable-auto-tool-choice
```

```bash
# vLLM serves an OpenAI-compatible API at http://localhost:8000/v1. Use temperature 1.0 and top_p 0.95,
# and send the model's earlier thinking back with each turn
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Qwen3-Embedding-8B | BB | 75.4 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-8b.min.md |
| Gemma 4 31B IT | BB | 74.1 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-31b-it.min.md |
| Qwen3-Embedding-4B | BB | 73.6 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-4b.min.md |
| Gemma 4 12B IT | BB | 72.3 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-12b-it.min.md |
| Gemma 4 26B A4B IT | BB | 72.2 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)