Skip to page content
SAVRN
Search Plan a project

Open-weight model · Text generation

Qwen3-4B-Q4_K_M-GGUF

by Shayon Sarker ShayonSarker/Qwen3-4B-Q4_K_M-GGUF

Qwen3-4B-Q4_K_M-GGUF is an open-weight model for text generation from Shayon Sarker, released under Apache License 2.0. Its published files total 6.8 GB. It draws 139 downloads a month.

Downloads: Hugging Face — ShayonSarker/Qwen3-4B-Q4KM-GGUF · Source & build recipes: GitHub — model-qwen3-4b-q4-k-m-gguf · Base model: Qwen/Qwen3-4B The GGUF binaries are hosted on Hugging Face; this repository holds the reproducible conversion pipeline…

Parameters—
Context—
Weights6.8 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads139

Model Card

By Shayon Sarker, published under apache-2.0, revision 95d37db2ae03.

Downloads: Hugging Face — ShayonSarker/Qwen3-4B-Q4KM-GGUF · Source & build recipes: GitHub — model-qwen3-4b-q4-k-m-gguf · Base model: Qwen/Qwen3-4B The GGUF binaries are hosted on Hugging Face; this repository holds the reproducible conversion pipeline, calibration data, and checksums. Base: Qwen/Qwen3-4B (Apache-2.0) · Quant: llama.cpp 0.4.1-dev (60081bb) · imatrix on 3000-line wikitext-2 (seed 42, 618 chunks, 4 threads) The F16 row in the eval table below is a measurement reference, not a downloadable file. Gate was <5% — passed. Task evals not run: perplexity only, stated honestly. Files: smol-Q4KM.gguf, smol-Q80.gguf, imatrix.dat, calib.txt, REPRO.json, SHA256.txt. Full repro: see…

Read Shayon Sarker's full model card

Qwen3-4B Q4_K_M GGUF — verified 4GB edge build

Downloads: Hugging Face — ShayonSarker/Qwen3-4B-Q4_K_M-GGUF · Source & build recipes: GitHub — model-qwen3-4b-q4-k-m-gguf · Base model: Qwen/Qwen3-4B

The GGUF binaries are hosted on Hugging Face; this repository holds the reproducible conversion pipeline, calibration data, and checksums.

Base: Qwen/Qwen3-4B (Apache-2.0) · Quant: llama.cpp 0.4.1-dev (60081bb) · imatrix on 3000-line wikitext-2 (seed 42, 618 chunks, 4 threads)

Format status

File Status Note
smol-Q4_K_M.gguf Published Main release, ~2.4 GB
smol-Q8_0.gguf Published ~4.0 GB
F16 Reference only 8.0 GB source-side reference used for the PPL baseline; not published

The F16 row in the eval table below is a measurement reference, not a downloadable file.

Eval (wikitext, 128 held-out lines, 2048 ctx)

model PPL size
F16 11.4970 8.0 GB
Q8_0 11.5480 (+0.44%) 4.0 GB
Q4_K_M 11.8856 (+3.38%) 2.4 GB

Gate was <5% — passed. Task evals not run: perplexity only, stated honestly. Files: smol-Q4_K_M.gguf, smol-Q8_0.gguf, imatrix.dat, calib.txt, REPRO.json, SHA256.txt. Full repro: see GitHub link below.

Verified outputs

Verbatim smol-Q4_K_M.gguf completions, --temp 0 -no-cnv, 24 new tokens, prompt form Question: ...\nAnswer:. Qwen3-4B ships a chat template, so without -no-cnv llama.cpp switches to conversation mode and ignores -p.

Question Model answer
What is the capital of Japan? The capital of Japan is Tokyo. It is the largest city in Japan and serves as the political, economic, and cultural
What is the capital of Italy? The capital of Italy is Rome. Rome is a city in the central part of Italy, and it is known for its
What is the capital of Egypt? The capital of Egypt is Cairo. It is a city that has been the political, economic, and cultural center of Egypt
What is the largest ocean on Earth? The Pacific Ocean is the largest ocean on Earth, covering approximately 63 million square miles (165 million square
Which planet is closest to the Sun? Mercury is the planet closest to the Sun. It is the smallest planet in our solar system and has the shortest orbital period
How many days are in a leap year? A leap year has 366 days. This is because, in addition to the usual 365 days
How many continents are there? There are seven continents: Africa, Antarctica, Asia, Europe, North America, Australia, and South America.
What is the chemical symbol for gold? The chemical symbol for gold is Au.

Eight of eight are correct. Sentences are cut at the 24-token limit, so the trailing text is incomplete by construction.

Use

Ollama (see Modelfile, non-thinking params per Qwen3 best practices):

ollama create qwen3-4b-q4 -f Modelfile
ollama run qwen3-4b-q4 "Explain overfitting in 3 lines."

llama.cpp:

./llama-cli -m smol-Q4_K_M.gguf -p "Explain overfitting in 3 lines." -c 4096

Q4 peaks ~3GB: fits 4GB GPUs, fully offline. For Qwen3 thinking mode use Temperature=0.6, TopP=0.95; for fast non-thinking answers use Temperature=0.7, TopP=0.8 (this Modelfile).

Limitations

  • Perplexity-only validation; verify on your task before production use.
  • English wikitext calibration: other languages/domains may vary.
  • Qwen3 can repeat itself under greedy decoding — use the sampling params above.

Identity and Version

Repository
ShayonSarker/Qwen3-4B-Q4_K_M-GGUF
Publisher
Shayon Sarker
Task
Text generation
Modality
Text
Library
Not stated by the source
Parameters
Not stated by the source
Languages
Not stated by the source
Revision
95d37db2ae03bf0d9dfc4e234ef299923157992e
First published
2026-09-19
Last updated
2026-09-26

Files and Weights

10 files, 6.8 GB in total. The weights are 2 files totalling 6.8 GB in gguf.

Weights2 files · 6.8 GB
Configuration1 file · 608 B
Documentation1 file · 4.8 KB
Other5 files · 5.3 MB
Repository1 file · 1.7 KB
Every file
FileTypeSizeSHA-256
smol-Q4_K_M.ggufWeights2.5 GB aa961d70f7fa
smol-Q8_0.ggufWeights4.3 GB d4bde3a852f8
REPRO.jsonConfiguration608 B —
README.mdDocumentation4.8 KB —
ModelfileOther112 B —
SHA256.txtOther402 B —
UPLOAD.txtOther125 B —
calib.txtOther1.4 MB —
imatrix.datOther3.9 MB 8bec1a7e5d0e
.gitattributesRepository1.7 KB —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
6.8 GB
Download from Shayon Sarker

Released by Shayon Sarker through its official repository on Hugging Face. Read the license.

Built From

Memory Requirements

PrecisionWeights in memory
As published6.8 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About Qwen3-4B-Q4_K_M-GGUF

Can I use Qwen3-4B-Q4_K_M-GGUF commercially?

Yes. Qwen3-4B-Q4_K_M-GGUF is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

Similar Models

Model · Text generation

opt-125m

AI at Meta

OPT was first introduced in Open Pre-trained Transformer Language Models and first released in metaseq's repository on May 3rd 2022 by Meta AI. Disclaimer: The team releasing OPT wrote an official model card, which is available in Appendix D of the paper. Content from this model card has been written by the Hugging Face team. To quote the first two paragraphs of the official paper OPT was predominantly pretrained with English text, but a small amount of non-English data is still present within the training corpus via CommonCrawl. The model was pretrained using a causal language modeling (CLM) objective. OPT belongs to the same family of decoder-only models like GPT-3. As such, it was…

Open weights other 2,048 tokens transformers

Model · Text generation

Ornith-1.5-9B-GGUF

Ornith

Chirp Chirp! We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement. Ornith-1.5 extends Ornith-1.0 (which was developed on top of Qwen3.5 and Gemma4 with additional continued pretraining, mid-training, and post-training) by expanding the self-improvement loop from scaffold and rollout optimization to jointly optimizing task generation, scaffold construction, and solution rollouts. Rather than relying on a fixed set of human-curated tasks and manually designed harnesses, Ornith-1.5 continuously generates new training tasks, discovers effective strategies for solving them, and improves the policy through reinforcement learning. For…

Open weights mit transformers

Model · Text generation

Ornith-1.5-35B-A3B-GGUF

Ornith

Chirp Chirp! We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement. Ornith-1.5 extends Ornith-1.0 (which was developed on top of Qwen3.5 and Gemma4 with additional continued pretraining, mid-training, and post-training) by expanding the self-improvement loop from scaffold and rollout optimization to jointly optimizing task generation, scaffold construction, and solution rollouts. Rather than relying on a fixed set of human-curated tasks and manually designed harnesses, Ornith-1.5 continuously generates new training tasks, discovers effective strategies for solving them, and improves the policy through reinforcement learning. For…

Open weights mit transformers

Model · Text generation

Ternary-Bonsai-2-27B-gguf

Prism ML

Full 27B-class reasoning in ternary transformer weights, for llama.cpp (CUDA, Metal, CPU) - \~5.9 GB language model (down from \~54 GB FP16) — full 27B-class reasoning on a standard laptop or a single GPU - 98.2% of FP16 intelligence retained: 84.78 average across 14 thinking-mode benchmarks — far above the conventional IQ2XXS build (72.59) at about 82% of its footprint, and within 0.4 points of UD-Q4KXL at three times the footprint - Retains thinking, reasoning, and agentic behavior deep in the sub-4-bit regime, where conventional low-bit representations collapse: math within half a point of full precision (96.57), coding level with the baseline (89.42), agentic tool calling at 74.92…

Open weights apache-2.0 llama.cpp

Uncensored Qwen3.8-27B, published as GGUF quantizations with the multi token prediction (MTP) head retained and verified. Refusal behaviour has been substantially reduced, not eliminated. See Measured behaviour for the numbers. Capabilities, training data, and architecture are otherwise unchanged. - Refusal directions removed with Heretic, which co minimizes refusal count against KL divergence from the base model. No handwritten refusal removal code, no finetuning, no additional training data. - Abliteration runs at bf16 (no 4 bit quantization). the resulting LoRA is merged into the bf16 base, so the published weights are not a quantized round trip. - mtp. tensors are copied verbatim from…

Open weights apache-2.0 llama.cpp

Model · Text generation

Ornith-1.0-9B-GGUF

Ornith

Aloha! Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding. This model card documents Ornith-1.0-9B, the most lightweight member of the Ornith family, designed for efficient single-GPU deployment. Ornith-1.0-9B is a dense ~9B model (≈19 GB in bf16), so it serves comfortably on a single 80GB GPU. The recipes below stand up an OpenAI-compatible server; add --tensor-parallel-size / --tp if you want to shard across more GPUs. For a quick local test (or to script offline generation), load the model directly with Transformers. Make sure you have a recent release installed — see the Transformers installation guide; Ornith-1.0-9B requires…

Open weights mit transformers