AIZYBRAIN-NANO-4B (GGUF)

AIZYBRAIN-NANO-4B is a high-performance, compact language model in GGUF format, specifically optimized for local inference. It is based on the Qwen3-4B architecture, designed to deliver a perfect balance between reasoning depth and computational efficiency.

🚀 Model Overview

• Developer: steef68
• Architecture: Qwen3 (Causal Language Model)
• Base Model: Qwen3-4B-Instruct
• Parameters: 4 Billion
• Format: GGUF
• Context Window: 32,768 tokens
• License: Apache 2.0

✨ Key Features

• Reasoning Capabilities: Native support for "Chain-of-Thought" processing using <think> blocks, allowing the model to solve complex logical problems before answering.
• Optimized for Edge Devices: Specifically tuned to run smoothly on consumer-grade hardware, including laptops with limited VRAM and CPU-only setups.
• Multilingual Expertise: Exceptional performance in French and English, with robust understanding across 20+ additional languages.
• High Efficiency: Utilizes Grouped-Query Attention (GQA) for faster inference and lower memory consumption.

🛠 Installation and Usage

For LM Studio / AnythingLLM

1. Search for steef68/AIZYBRAIN-NANO-4B in the app.
2. Download the AIZYBRAIN-V2-4B.gguf file.
3. Select the ChatML or Qwen prompt template.

For Ollama

You can use this model by creating a Modelfile:

FROM ./AIZYBRAIN-V2-4B.gguf
PARAMETER temperature 0.7
TEMPLATE "{{ if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{ if .Prompt }}<|im_start|>user
{{ .Prompt }}<|im_end|>
{{ end }}<|im_start|>assistant
"