Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

Thinks through problems step by step inside think tags, as a 4B Qwen3.5 model fine-tuned on reasoning chains and packaged as GGUF. For people who run a small reasoning model on their own computer.

Made by Jackrong on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This is a GGUF version of a small reasoning model built on Qwen3.5-4B. The author fine-tuned it on reasoning chains that came mainly from Claude-4.6 Opus interactions, so it plans in steps inside think tags before it answers. It handles text only.

Who it suits: People who want a small model that shows its reasoning and runs on their own computer, and learners who want to see how such fine-tuning is done.

Industry | Who uses it | How they use it | Example
Education | Students | Ask for step-by-step working on math or science questions | A physics problem solved with the steps laid out
Software | Developers | Run a small local model for planning and analysis | A task broken into parts before any code is written
Research | Hobbyists | Study how reasoning distillation changes a small model | Compare its thinking with the base model
Writing | Analysts | Think through an argument before drafting | A list of constraints and edge cases for a plan

Skip it if you need image input, because the final model is text only.

Where to use it:

• Your own computer, on Windows, macOS or Linux, once Python has fetched the files
• Local runners that open GGUF files

Use case | Who | What they get back
Step-by-step problem solving | Students | A reasoning trace in think tags, then an answer
Task planning | Developers | A plan that lists the goal, parts, constraints and checks
Training study | Learners | A model whose fine-tuning the author documented

How to set it up: You need Python with pip, a local runner for GGUF files and about 2.1 GB of disk at 4-bit by the record's estimate.

• Python and pip
• A runner that opens GGUF files
• Around 2.1 GB of disk space for a 4-bit file

1. Install the Hugging Face hub tool.
2. Download the repository's files.
3. Open a GGUF file in your runner.

pip install -U huggingface_hub
hf download Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

How to use it: Chat with it in your runner and give it problems that benefit from steps. The model writes its thinking first, inside think tags, then gives the final solution.

hf download Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

Tool | What it does
hf download | Copies this repository's files to your computer
think tags | Hold the model's step-by-step reasoning before the answer

Example: You ask it to plan how to solve a word problem. It first writes a short analysis, naming the goal, splitting the task, checking constraints and edge cases, then it gives a solution.

For newcomers and experts: If you are new to this, load the file and ask one math question. Watch the thinking section before the answer.

If you know the field, the author trained with supervised fine-tuning and LoRA, masking the loss to the assistant's thinking and reply. The card says later data from a larger Qwen3.5 model added science, instruction-following and math reasoning. It aims to cut repetitive loops on simple questions.

Start from Qwen3.5-4B
Supervised fine-tuning with LoRA
Train on structured reasoning chains
Package the result as GGUF

Good to know: The card makes claims about efficiency that this page cannot confirm. Test the model on your own questions before you rely on it.

• Languages: English, Chinese and Korean are listed
• Training material: the author published a notebook, code and a guide for the fine-tuning process
• License: Apache 2.0
• Cost to run: none named
• Data it reaches: the files you download from Hugging Face, plus the text you give your local runner
• Not for: image input, since the final model is text only

On Hugging Face, 13 November 2026

• Task: text generation
• Parameters: 4.2B
• Weights: about 2.1 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 39,987
• Likes: 188
• License: apache-2.0
• Languages: en, zh, ko
• Based on: Qwen/Qwen3.5-4B
• Tags: gguf, qwen3_5, unsloth, qwen, qwen3.5, reasoning, chain-of-thought, lora, text-generation, conversational

Links

• Model: https://huggingface.co/Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF
• Files: https://huggingface.co/Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.