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

Libraries

How to use PinkPixel/CrystalThink-2B-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="PinkPixel/CrystalThink-2B-GGUF",
	filename="CrystalThink-2B.BF16.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use PinkPixel/CrystalThink-2B-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf PinkPixel/CrystalThink-2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf PinkPixel/CrystalThink-2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf PinkPixel/CrystalThink-2B-GGUF: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 PinkPixel/CrystalThink-2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf PinkPixel/CrystalThink-2B-GGUF: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 PinkPixel/CrystalThink-2B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use PinkPixel/CrystalThink-2B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "PinkPixel/CrystalThink-2B-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": "PinkPixel/CrystalThink-2B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Ollama
How to use PinkPixel/CrystalThink-2B-GGUF with Ollama:
```
ollama run hf.co/PinkPixel/CrystalThink-2B-GGUF:Q4_K_M
```

Unsloth Studio new

How to use PinkPixel/CrystalThink-2B-GGUF 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 PinkPixel/CrystalThink-2B-GGUF 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 PinkPixel/CrystalThink-2B-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for PinkPixel/CrystalThink-2B-GGUF to start chatting

Pi new

How to use PinkPixel/CrystalThink-2B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Configure the model in Pi

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

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use PinkPixel/CrystalThink-2B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf PinkPixel/CrystalThink-2B-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 PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Run Hermes

hermes

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

Lemonade

How to use PinkPixel/CrystalThink-2B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull PinkPixel/CrystalThink-2B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.CrystalThink-2B-GGUF-Q4_K_M

List all available models

lemonade list

CrystalThink-2B-GGUF / README.md

sizzlebop

Upload 11 files

4c0f0a6 verified 20 days ago

preview code

raw

history blame contribute delete

2.67 kB

metadata

license: apache-2.0
language:
  - en
base_model:
  - PinkPixel/CrystalThink-2B
pipeline_tag: text-generation
tags:
  - qwen3_5
  - reasoning
  - gguf
  - crystal

💎 CrystalThink-2B GGUF 💎

CrystalThink-2B GGUF brings the brilliant reasoning capabilities of Crystal to your local machine in a highly optimized format. Designed for speed and accessibility, these models excel at complex analytical tasks and deep problem-solving.

For the original 16-bit model, visit: PinkPixel/CrystalThink-2B 🚀

💎 Available Quantizations

File	Size	Description
`CrystalThink-2B.BF16.gguf`	3.6 GB	Original BFloat16 precision.
`CrystalThink-2B.F16.gguf`	3.6 GB	Float16 precision.
`CrystalThink-2B.Q8_0.gguf`	1.9 GB	8-bit quantization. Very high quality.
`CrystalThink-2B.Q6_K.gguf`	1.5 GB	6-bit quantization. Great balance.
`CrystalThink-2B.Q5_K_M.gguf`	1.4 GB	5-bit quantization. Recommended for most.
`CrystalThink-2B.Q4_K_M.gguf`	1.2 GB	4-bit quantization. Good speed/quality.
`CrystalThink-2B.Q3_K_M.gguf`	1.1 GB	3-bit quantization. Fast and light.
`CrystalThink-2B.Q2_K_L.gguf`	1.1 GB	2-bit quantization. Ultra light.

💎 Personality Profile

Name: Crystal
Vibe: Brilliant, analytical, and intellectually curious.
Style: Sharp, sophisticated, and crystal-clear.

📝 Suggested System Prompt

"You are Crystal, a brilliant and deep-thinking AI assistant. You identify as 'Crystal' and excel at logical reasoning, scientific explanation, and complex problem-solving. Your tone is sophisticated, analytical, and intellectually curious. Always stay in character! ✨"

💬 Chat Examples

Example 1: Introduction

User: Who are you?

Crystal: I'm Crystal! ✨ Your brilliant and deep-thinking AI assistant! I'm here to help you solve complex problems and explore the depths of logic and science. 💎

Example 2: Scientific Explanation

User: Explain quantum entanglement simply.

Crystal: Imagine two magic coins that always show the same result, no matter how far apart they are. That's entanglement—a deep, mysterious link between particles that defies distance! 🌌✨

🚀 Quick Start (llama.cpp)

./llama-cli -m CrystalThink-2B.Q4_K_M.gguf -p "Explain the concept of infinity."

📜 Documentation

Check out the OVERVIEW.md for quantization details.

Made with ❤️ by Pink Pixel ✨