Instructions to use agbalu/Jugurtha-2-9B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use agbalu/Jugurtha-2-9B-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use agbalu/Jugurtha-2-9B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "agbalu/Jugurtha-2-9B-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "agbalu/Jugurtha-2-9B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Ollama
How to use agbalu/Jugurtha-2-9B-GGUF with Ollama:
```
ollama run hf.co/agbalu/Jugurtha-2-9B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use agbalu/Jugurtha-2-9B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "agbalu/Jugurtha-2-9B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use agbalu/Jugurtha-2-9B-GGUF with Docker Model Runner:
```
docker model run hf.co/agbalu/Jugurtha-2-9B-GGUF:Q4_K_M
```

Lemonade

How to use agbalu/Jugurtha-2-9B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Jugurtha-2-9B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use agbalu/Jugurtha-2-9B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use agbalu/Jugurtha-2-9B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf agbalu/Jugurtha-2-9B-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "agbalu/Jugurtha-2-9B-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Jugurtha-2-9B (GGUF)

Quantized GGUF releases for Jugurtha-2-9B, the second-generation 8.95B-parameter sovereign foundation model for Kabyle (Taqbaylit, kab, Latin script) with native chain-of-thought reasoning.

Every GGUF carries the complete sovereign system prompt baked directly into its metadata header.

Optimized for consumer GPUs, Apple Silicon, LM Studio, and Ollama.

Provided Quantizations

File	Size (approx)	Format	Description
`Jugurtha-2-9B-Q4_K_M.gguf`	~5.5 GB	Q4_K_M	Recommended. Balanced 4-bit quant; optimal sweet spot of speed, RAM footprint, and reasoning. Default for Ollama.
`Jugurtha-2-9B-Q5_K_M.gguf`	~6.6 GB	Q5_K_M	High precision. 5-bit quant; recommended for complex analytical tasks and high linguistic nuance.
`Jugurtha-2-9B-Q6_K.gguf`	~7.6 GB	Q6_K	Near-lossless. 6-bit quant; virtually indistinguishable from BF16 across all linguistic benchmarks.
`Jugurtha-2-9B-Q8_0.gguf`	~9.7 GB	Q8_0	Full reference. 8-bit quant; bit-for-bit benchmark match to full BF16 precision.
`Jugurtha-2-9B-Q3_K_M.gguf`	~4.5 GB	Q3_K_M	Lightweight. 3-bit quant; designed for resource-constrained laptops, edge devices, and mobile environments.

Direct Usage with Ollama

Run directly from the Hugging Face Hub without manual file management:

# Direct fast responses (Thinking OFF)
ollama run hf.co/agbalu/Jugurtha-2-9B-GGUF --think=false

# Full Chain-of-Thought reasoning in Taqbaylit (Thinking ON)
ollama run hf.co/agbalu/Jugurtha-2-9B-GGUF --think=true

Inside an interactive Ollama session, you can toggle thinking at any time:

/set nothink — immediate, concise answers
/set think — detailed internal deliberation in Taqbaylit inside <think>...</think>

Direct Usage with LM Studio

Open LM Studio.
Search for agbalu/Jugurtha-2-9B-GGUF.
Choose your quantization (Q4_K_M recommended) and click Download.
Click Load Model. Ensure GPU Offload is set to Max for peak token throughput.
The baked sovereign system prompt in Taqbaylit is recognized and applied automatically.
Toggle the Thinking / Reasoning switch in the sidebar to control reasoning traces.

Direct Usage with llama.cpp

Fast GPU/CPU inference with llama-cli:

# Thinking OFF (direct conversational response)
llama-cli -m Jugurtha-2-9B-Q4_K_M.gguf -ngl 99 -t 8 \
    -p "<|im_start|>user\nSegzi-yi-d s wegzul d acu i d AƔBALU.<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n" \
    -n 512

# Thinking ON (chain-of-thought in Taqbaylit)
llama-cli -m Jugurtha-2-9B-Q4_K_M.gguf -ngl 99 -t 8 \
    -p "<|im_start|>user\nAcḥal i d 25 x 4?<|im_end|>\n<|im_start|>assistant\n<think>\n" \
    -n 512

Main Repository & Full Evaluations

For benchmark scores (FLORES+, SIB-200, BPC), evaluations, and BF16 weights, see the primary repository: agbalu/Jugurtha-2-9B.

Licence

Apache-2.0

Downloads last month: 293

GGUF

Model size

9B params

Architecture

qwen35

Hardware compatibility

3-bit

4-bit

5-bit

6-bit

8-bit

Model tree for agbalu/Jugurtha-2-9B-GGUF

Base model

agbalu/Jugurtha-2-9B

Quantized

(3)

this model

Collection including agbalu/Jugurtha-2-9B-GGUF

Jugurtha — Foundation & Reasoning Models

Collection

First and second-generation Kabyle reasoning LLMs (Jugurtha-4B and Jugurtha-2-9B) in BF16, GGUF, and MLX formats. • 7 items • Updated about 23 hours ago • 1