Instructions to use CodexCon-OS/Amethyst-Core with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use CodexCon-OS/Amethyst-Core with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="CodexCon-OS/Amethyst-Core",
	filename="amethyst-arc-1b-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use CodexCon-OS/Amethyst-Core 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 CodexCon-OS/Amethyst-Core:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf CodexCon-OS/Amethyst-Core:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf CodexCon-OS/Amethyst-Core:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf CodexCon-OS/Amethyst-Core:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf CodexCon-OS/Amethyst-Core:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf CodexCon-OS/Amethyst-Core:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf CodexCon-OS/Amethyst-Core:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf CodexCon-OS/Amethyst-Core:Q4_K_M

Use Docker

docker model run hf.co/CodexCon-OS/Amethyst-Core:Q4_K_M

LM Studio
Jan

vLLM

How to use CodexCon-OS/Amethyst-Core with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/CodexCon-OS/Amethyst-Core:Q4_K_M

Ollama
How to use CodexCon-OS/Amethyst-Core with Ollama:
```
ollama run hf.co/CodexCon-OS/Amethyst-Core:Q4_K_M
```

Unsloth Studio

How to use CodexCon-OS/Amethyst-Core with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for CodexCon-OS/Amethyst-Core to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for CodexCon-OS/Amethyst-Core to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for CodexCon-OS/Amethyst-Core to start chatting

Atomic Chat new
Docker Model Runner
How to use CodexCon-OS/Amethyst-Core with Docker Model Runner:
```
docker model run hf.co/CodexCon-OS/Amethyst-Core:Q4_K_M
```

Lemonade

How to use CodexCon-OS/Amethyst-Core with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull CodexCon-OS/Amethyst-Core:Q4_K_M

Run and chat with the model

lemonade run user.Amethyst-Core-Q4_K_M

List all available models

lemonade list

🔮 XLPHY Amethyst (Gemma Series) for Project: XLPHY AI

XLPHY Amethyst is a suite of high-efficiency, local-first AI models optimized specifically for the Project: XLPHY AI ecosystem. These models are repackaged and quantized to provide a premium, low-latency, and multimodal experience for autonomous agents and sovereign AI applications.

Developer Note: These are optimized derivatives of the Google Gemma 3 series, rebranded and tuned for seamless integration within the Project: XLPHY AI autonomous agent architecture.

🧠 Model Selection

The Amethyst series is built for Project: XLPHY AI and is divided into four "Gemstone Tiers." Each tier is available in Q4_K_M, Q5_K_M, and Q6_K quantization levels.

File Name (Template)	Tier Identity	Base Engine	Primary Purpose	License	Available Quants
`amethyst-arc-1b-[quant].gguf`	arc	Gemma 3 1B IT	Ultra-fast local execution and IoT.	Gemma	`Q4_K_M`
`amethyst-beam-e2b-[quant].gguf`	Core	Gemma 4 E2B IT	Main driver with Vision support.	Apache 2.0	`Q4_K_M`, `Q6_K`, `Q8_0`

📦 Quantization Guide

Q4_K_M (Low): Fastest and most memory-efficient. Ideal for mobile and entry-level hardware.
Q5_K_M (Medium): The "sweet spot" for Amethyst, with minimal quality differences from the original model.
Q6_K (High): Near-lossless performance for users who prioritize maximum accuracy.

🛠️ Implementation & Runtime

Designed for the Project: XLPHY AI "Offline-First" philosophy. Best executed via:

XLPHY Desktop App (Native Integration)
llama.cpp / llama-cli
Any GGUF-compatible inference engine supporting Gemma 3 and Gemma 4

🔐 Checksums (SHA256)

To ensure file integrity during the XLPHY automated download process:

Tier	Q4_K_M	Q6_K	Q8_0	mmproj
Arc (1B)	`12bf0fff8815d5f73a3c9b586bd8fee8e7b248c935de70dec367679873d0f29d`	`X`	`X`	`X`
Beam (E2B)	`6c950d754366dd8b372fd17a40497ba5f130a46d833b4c5bccc9f6bb6382ce1e`	`5a352f63b59e1b37bac5071e3aed65e8954883a1625bf6676bf9aa2381295beb`	`6febf73c7c6b44c550ba857ceca3788262ce2272603ca16b1f2609556205dea3`	`b23366bec60eda708aa54816c8f72dfdfedd5c5ccbe4e7db500aa7724bd14b20`

⚖️ Attribution & Licensing

These files are redistributed/repackaged quantized derivatives of the Google Gemma family.

Original Architecture: Developed by Google DeepMind
Optimization: Repackaged by CodexCon Digital Solutions for Project: XLPHY AI
Amethyst Arc (Gemma 3): Gemma Terms of Use
Amethyst Beam (Gemma 4): Apache License 2.0

⚠️ Limitations & Safety

Hallucinations: Like all LLMs, these models may produce incorrect information.
Human-in-the-loop: Always validate technical outputs, especially for vision-based tasks or critical code.
Non-Critical Use: Not intended for medical, legal, or other high-stakes safety-critical applications.

Developed by CodexCon | Lead Founder: Cid Cruz

Downloads last month: 198

GGUF

Model size

1.0B params

Architecture

gemma3

Hardware compatibility

4-bit

6-bit

8-bit

Model tree for CodexCon-OS/Amethyst-Core

Base model

google/gemma-3-1b-pt

Finetuned

google/gemma-3-1b-it

Quantized

(449)

this model