Qwen_Qwen3-4B-GGUF Open-source Model - A Powerful Tool for Text Generation Supporting Multiple Quantization Types

Qwen Qwen3 4B GGUF

Developed by bartowski

The Llamacpp imatrix quantization version of Qwen3-4B provided by the Qwen team, supporting multiple quantization types and suitable for text generation tasks.

Large Language Model #Multiple Quantization Version Selection #Low Memory Optimization #Chinese Text Generation

Downloads 10.58k

Release Time : 4/28/2025

Model Overview

A quantized version based on Qwen/Qwen3-4B, using llama.cpp for quantization, and can be run in LM Studio or llama.cpp and its derivative projects.

Model Features

Multiple Quantization Options

Offers various quantization types from Q2_K to Q8_0 to meet different hardware and performance needs.

imatrix Quantization

Uses the imatrix option for quantization to enhance model performance.

Supports LM Studio and llama.cpp

Can be run in LM Studio or directly using llama.cpp and its derivative projects.

Embedding/Output Weight Optimization

Some quantized versions quantize the embedding and output layer weights to Q8_0 to improve model quality.

Model Capabilities

Text Generation

Multi-turn Dialogue

Supports System Prompts

Use Cases

Dialogue Systems

Intelligent Assistant

Used to build intelligent assistants that support multi-turn dialogues and system prompts.

Text Generation

Content Creation

Used to generate articles, stories, and other textual content.

🚀 Llamacpp imatrix Quantizations of Qwen3-4B by Qwen

This project provides quantized versions of the Qwen3-4B model using the llama.cpp library. It offers various quantization types to balance between model size and performance, making it suitable for different hardware configurations.

🚀 Quick Start

Prerequisites

Ensure you have llama.cpp installed. You can download it from here.
For downloading models, you may need the huggingface-cli. Install it using pip install -U "huggingface_hub[cli]".

Running the Model

You can run the quantized models in LM Studio.
Alternatively, use llama.cpp or any other llama.cpp-based project to run the models directly.

✨ Features

Multiple Quantization Types: Offers a wide range of quantization types, such as bf16, Q8_0, Q6_K_L, etc., to meet different performance and memory requirements.
Online Repacking: Some quantization types support online repacking, which can improve performance on ARM and AVX machines.
Embed/Output Weights Optimization: Certain quantizations use Q8_0 for embeddings and output weights to enhance quality.

📦 Installation

Using huggingface-cli

Click to view download instructions

First, make sure you have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download bartowski/Qwen_Qwen3-4B-GGUF --include "Qwen_Qwen3-4B-Q4_K_M.gguf" --local-dir ./

If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download bartowski/Qwen_Qwen3-4B-GGUF --include "Qwen_Qwen3-4B-Q8_0/*" --local-dir ./

You can either specify a new local-dir (Qwen_Qwen3-4B-Q8_0) or download them all in place (./)

💻 Usage Examples

Prompt Format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

📚 Documentation

Download a File

Filename	Quant type	File Size	Split	Description
Qwen3-4B-bf16.gguf	bf16	8.05GB	false	Full BF16 weights.
Qwen3-4B-Q8_0.gguf	Q8_0	4.28GB	false	Extremely high quality, generally unneeded but max available quant.
Qwen3-4B-Q6_K_L.gguf	Q6_K_L	3.40GB	false	Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended.
Qwen3-4B-Q6_K.gguf	Q6_K	3.31GB	false	Very high quality, near perfect, recommended.
Qwen3-4B-Q5_K_L.gguf	Q5_K_L	2.98GB	false	Uses Q8_0 for embed and output weights. High quality, recommended.
Qwen3-4B-Q5_K_M.gguf	Q5_K_M	2.89GB	false	High quality, recommended.
Qwen3-4B-Q5_K_S.gguf	Q5_K_S	2.82GB	false	High quality, recommended.
Qwen3-4B-Q4_1.gguf	Q4_1	2.60GB	false	Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon.
Qwen3-4B-Q4_K_L.gguf	Q4_K_L	2.59GB	false	Uses Q8_0 for embed and output weights. Good quality, recommended.
Qwen3-4B-Q4_K_M.gguf	Q4_K_M	2.50GB	false	Good quality, default size for most use cases, recommended.
Qwen3-4B-Q4_K_S.gguf	Q4_K_S	2.38GB	false	Slightly lower quality with more space savings, recommended.
Qwen3-4B-Q4_0.gguf	Q4_0	2.38GB	false	Legacy format, offers online repacking for ARM and AVX CPU inference.
Qwen3-4B-IQ4_NL.gguf	IQ4_NL	2.38GB	false	Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference.
Qwen3-4B-Q3_K_XL.gguf	Q3_K_XL	2.33GB	false	Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability.
Qwen3-4B-IQ4_XS.gguf	IQ4_XS	2.27GB	false	Decent quality, smaller than Q4_K_S with similar performance, recommended.
Qwen3-4B-Q3_K_L.gguf	Q3_K_L	2.24GB	false	Lower quality but usable, good for low RAM availability.
Qwen3-4B-Q3_K_M.gguf	Q3_K_M	2.08GB	false	Low quality.
Qwen3-4B-IQ3_M.gguf	IQ3_M	1.96GB	false	Medium-low quality, new method with decent performance comparable to Q3_K_M.
Qwen3-4B-Q3_K_S.gguf	Q3_K_S	1.89GB	false	Low quality, not recommended.
Qwen3-4B-IQ3_XS.gguf	IQ3_XS	1.81GB	false	Lower quality, new method with decent performance, slightly better than Q3_K_S.
Qwen3-4B-Q2_K_L.gguf	Q2_K_L	1.76GB	false	Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable.
Qwen3-4B-IQ3_XXS.gguf	IQ3_XXS	1.67GB	false	Lower quality, new method with decent performance, comparable to Q3 quants.
Qwen3-4B-Q2_K.gguf	Q2_K	1.67GB	false	Very low quality but surprisingly usable.
Qwen3-4B-IQ2_M.gguf	IQ2_M	1.51GB	false	Relatively low quality, uses SOTA techniques to be surprisingly usable.

Embed/Output Weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

ARM/AVX Information

Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass.

Now, however, there is something called "online repacking" for weights. details in this PR. If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.

As of llama.cpp build b4282 you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0.

Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to this PR which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase.

Click to view Q4_0_X_X information (deprecated

I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking.

Click to view benchmarks on an AVX2 system (EPYC7702)

model	size	params	backend	threads	test	t/s	% (vs Q4_0)
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	pp512	204.03 ± 1.03	100%
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	pp1024	282.92 ± 0.19	100%
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	pp2048	259.49 ± 0.44	100%
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	tg128	39.12 ± 0.27	100%
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	tg256	39.31 ± 0.69	100%
qwen2 3B Q4_0	1.70 GiB	3.09 B	CPU	64	tg512	40.52 ± 0.03	100%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	pp512	301.02 ± 1.74	147%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	pp1024	287.23 ± 0.20	101%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	pp2048	262.77 ± 1.81	101%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	tg128	18.80 ± 0.99	48%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	tg256	24.46 ± 3.04	83%
qwen2 3B Q4_K_M	1.79 GiB	3.09 B	CPU	64	tg512	36.32 ± 3.59	90%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	pp512	271.71 ± 3.53	133%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	pp1024	279.86 ± 45.63	100%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	pp2048	320.77 ± 5.00	124%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	tg128	43.51 ± 0.05	111%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	tg256	43.35 ± 0.09	110%
qwen2 3B Q4_0_8_8	1.69 GiB	3.09 B	CPU	64	tg512	42.60 ± 0.31	105%

Q4_0_8_8 offers a nice bump to prompt processing and a small bump to text generation

Which File Should I Choose?

Click here for details

A great write up with charts showing various performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.

Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.

If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.

If you want to get more into the weeds, you can check out this extremely useful feature chart:

llama.cpp feature matrix

But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.

These I-quants can also be used on CPU, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.

🔧 Technical Details

Quantization Method

Using llama.cpp release b5200 for quantization. All quants are made using the imatrix option with a dataset from here.

📄 License

No license information provided in the original README.

Credits

Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset. Thank you ZeroWw for the inspiration to experiment with embed/output. Thank you to LM Studio for sponsoring my work.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご