Instructions to use ZeroXClem/Qwen3-4B-ChromaticCoder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ZeroXClem/Qwen3-4B-ChromaticCoder with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ZeroXClem/Qwen3-4B-ChromaticCoder")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ZeroXClem/Qwen3-4B-ChromaticCoder")
model = AutoModelForCausalLM.from_pretrained("ZeroXClem/Qwen3-4B-ChromaticCoder")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ZeroXClem/Qwen3-4B-ChromaticCoder with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ZeroXClem/Qwen3-4B-ChromaticCoder"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroXClem/Qwen3-4B-ChromaticCoder",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ZeroXClem/Qwen3-4B-ChromaticCoder

SGLang

How to use ZeroXClem/Qwen3-4B-ChromaticCoder with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ZeroXClem/Qwen3-4B-ChromaticCoder" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroXClem/Qwen3-4B-ChromaticCoder",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ZeroXClem/Qwen3-4B-ChromaticCoder" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroXClem/Qwen3-4B-ChromaticCoder",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ZeroXClem/Qwen3-4B-ChromaticCoder with Docker Model Runner:
```
docker model run hf.co/ZeroXClem/Qwen3-4B-ChromaticCoder
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

ZeroXClem/Qwen3-4B-ChromaticCoder

ZeroXClem/Qwen3-4B-ChromaticCoder is a vibrant and versatile 4B model fusion built using MergeKit and the model_stock strategy. Blending deep reasoning, mathematical precision, frontend UI generation, and code synthesis, it shines in logic-driven and creative problem spaces.

This model is a chromatic cascade of top-performing Qwen3 derivatives and fine-tuned reasoning specialists — harmonizing technical accuracy with structured expressiveness across a wide domain of tasks.

🧠 Overview

ChromaticCoder is based on the powerful foundation of prithivMLmods/Lacaille-MoT-4B-Supreme2, integrating a spectrum of expert finetunes to produce a model specialized in:

📊 Mathematical and logical reasoning
💻 Frontend & UI code generation
🧮 Multi-step algorithmic thinking
🛠️ Code reasoning, explanation, and synthesis
📐 Structured technical content creation

🧬 Merge Details

Detail	Value
Merge Method	`model_stock`
Base Model	`prithivMLmods/Lacaille-MoT-4B-Supreme2`
Dtype	`bfloat16`
Tokenizer Source	`prithivMLmods/Lacaille-MoT-4B-Supreme2`

🧩 Models Merged

Menlo/Jan-nano – Agentic research-aligned model with MCP support.
prithivMLmods/Octans-Qwen3-UI-Code-4B – UI code generation with Tailwind/React.
prithivMLmods/Logics-Qwen3-Math-4B – Advanced math and logic reasoning.
prithivMLmods/Carinae-Qwen3-Radiation-4B – Balanced probabilistic modeling with multilingual reasoning.
prithivMLmods/Kepler-Qwen3-4B-Super-Thinking – Hybrid symbolic-probabilistic thought.
prithivMLmods/Bootes-Qwen3_Coder-Reasoning – Instruction-tuned code synthesis and stepwise debugging.
Loom-Labs/Apollo-1-4B – General-purpose reasoning and multilingual instruction following.
GetSoloTech/Qwen3-Code-Reasoning-4B – Competitive programming and reasoning powerhouse.

🌈 Chromatic Features

✨ Unified Expert Reasoning
Brings together multiple specialized reasoning modules — from UI generation to symbolic math and programming logic — into one coherent architecture.

🧠 Deep Logic and Event Simulation
Excels in modeling probabilistic systems, structured math, and algorithmic solutions with step-by-step clarity.

💻 Frontend & UI Coding Mastery
With Octans and Jan-nano integrations, this model generates accurate and readable frontend code (React, Tailwind, HTML5).

🧪 STEM-Specialized Performance
Fine-tuned on math, logic, and scientific problem domains, ChromaticCoder is a strong match for educational and research applications.

🛠️ Developer-Centric Reasoning
Instruction-tuned layers optimize code completion, refactoring, and explanation across Python, JS, C++, and more.

🌍 Multilingual Capabilities
Thanks to Apollo and Carinae, it supports over 80 languages in both reasoning and coding domains.

🔧 MergeKit Configuration

name: ZeroXClem-Qwen3-4B-ChromaticCoder
base_model: prithivMLmods/Lacaille-MoT-4B-Supreme2
dtype: bfloat16
merge_method: model_stock
models:
  - model: Menlo/Jan-nano
  - model: prithivMLmods/Octans-Qwen3-UI-Code-4B
  - model: prithivMLmods/Logics-Qwen3-Math-4B
  - model: prithivMLmods/Carinae-Qwen3-Radiation-4B
  - model: prithivMLmods/Kepler-Qwen3-4B-Super-Thinking
  - model: prithivMLmods/Bootes-Qwen3_Coder-Reasoning
  - model: Loom-Labs/Apollo-1-4B
  - model: GetSoloTech/Qwen3-Code-Reasoning-4B
tokenizer_source: prithivMLmods/Lacaille-MoT-4B-Supreme2

💡 Use Cases

🎓 STEM Tutoring & Education
🧮 Mathematical and Logical Explanation
🖥️ Frontend Development & Prototyping
📘 Technical Documentation
🧑‍💻 Algorithm Debugging & Refactoring
🤖 Agentic Reasoning and Simulated Tool Use

🧪 Limitations

Limited by 4B parameter size — may struggle with extremely long or open-domain contexts.
Some outputs may be verbose or over-explained depending on the base tuning weights.
Not suitable for unrestricted creative or emotional writing tasks.