---
tags:
- gguf
- llama.cpp
- unsloth
- vision-language-model
base_model:
- Qwen/Qwen3.5-4B
language:
- en
pipeline_tag: text-generation
---

# Qwable-4B-Distilled-GGUF

This model is a distilled version of **Qwen3.5-4B**, fine-tuned on scraped Claude Code sessions. It is designed to enhance reasoning and agentic capabilities while maintaining the efficiency of the 4B parameter scale.

## Model Overview
- **Architecture:** Qwen3.5-4B (Base)
- **Methodology:** Distillation
- **Dataset:** Scraped Claude Code sessions (Sensitive info redacted)
- **Training Framework:** Unsloth

## Intended Use
This model is intended for research and experimentation in agentic workflows and complex reasoning tasks. It demonstrates improved performance over the base Qwen3.5 model in specific code-centric scenarios.

### ⚠️ Beta Status & Limitations
**This model is currently in beta.**
- **Production Usage:** Not recommended for production environments.
- **Reliability:** Performance may vary; it has not undergone extensive safety or robustness testing.
- **Expectations:** Treat outputs as experimental and verify critical reasoning steps.