Qwen3.8-27B-CodeForge-3xMerge i1 GGUF

Qwen3.8-27B-CodeForge-3xMerge i1 GGUF

Importance-matrix GGUF quants of the Qwen3.8-27B-CodeForge-3xMerge model, from about 11 GB to 22.5 GB, for local use. For people running 27B models on their own computer.

Made by mradermacher 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 repository holds GGUF files of the Qwen3.8-27B-CodeForge-3xMerge model, quantized with an importance matrix (the i1 series). The files range from about 11 GB to 22.5 GB so you can trade quality for memory. The card says little about the model itself, which is a merge listed as based on another repository by xtools-at.

Who it suits: It suits people who run models locally with GGUF tools and want a 27B model in a size that fits their memory.

Industry | Who uses it | How they use it | Example
Software | Local model users | Run a large merged model on their own machine | Choosing the Q4_K_M file to fit memory
Research | Quantization testers | Compare importance-matrix quants | Testing a 3-bit file against a 4-bit one
Hobbyists | Tinkerers | Try merged models offline | Chatting with the model without a cloud service

You can skip it if you want details on what the model does best, since the card gives none. A separate static-quant repository also exists.

Where to use it: GGUF files load in GGUF-compatible tools on Windows, macOS or Linux. The card calls this a vision model and says its mmproj files, if any, live in the maker's static repository.

• GGUF-compatible local inference tools
• Windows, macOS or Linux
• The static repository, for any mmproj files

Use case | Who | What they get back
Local text chat | Hobbyists | A 27B model running on your own computer
Smaller footprint | Users with less memory | Files from about 11 GB upward
Imatrix experiments | Quantizers | An imatrix file for making your own quants

How to set it up: You need Python for the download tool and disk space for the file you pick. The 4-bit estimate is about 13.7 GB. The card points to a third-party guide for GGUF basics, including how to join multi-part files.

• Python and the Hugging Face tool
• Disk space for the chosen file
• A GGUF-compatible tool to load it

1. Choose a quant by size from the card's table.
2. Download that file, or the whole repository.
3. Load the file in your GGUF tool.

pip install -U huggingface_hub
hf download mradermacher/Qwen3.8-27B-CodeForge-3xMerge-i1-GGUF

How to use it: Load the chosen file in your GGUF-compatible tool and chat as you would with any model. The maker says IQ quants are often better than similar-sized non-IQ ones, and that the sizes are sorted by size, not by quality.

Quant | Size in GB | Note
i1-Q2_K | 11.0 | Smallest here, lower quality
i1-IQ3_S | 12.7 | Better than the Q3_K types
i1-Q4_K_S | 15.9 | Balanced size, speed and quality
i1-Q4_K_M | 16.9 | Fast and the maker's pick
i1-Q6_K | 22.5 | Close to the static Q6_K

Example: You download the Q4_K_M file at 16.9 GB and load it in a GGUF tool. You then type a question and the model answers from your own machine, with no data sent out.

For newcomers and experts: If you are new, start with the Q4_K_M file, which the maker marks as fast and recommended. Bigger files cost more memory and give better quality.

If you know the field, these are weighted imatrix quants, and an imatrix file is included for building your own. The maker says i1-Q6_K is practically like the static Q6_K. The base is a three-way merge, though the card does not describe how it was made.

Merged 27B base model
Importance matrix computed
Quantized to many sizes
You pick a file for your memory

Good to know: The card gives no benchmark scores for this model.

• License: Apache 2.0
• Cost to run: no API key, account or paid service is named
• Data it reaches: files downloaded from Hugging Face, and whatever you type into your local tool
• Not for: claims about quality, since the card gives no tests

Title: File size in GB
i1-Q2_K: 11.0 GB
i1-IQ3_S: 12.7 GB
i1-Q4_K_S: 15.9 GB
i1-Q4_K_M: 16.9 GB
i1-Q6_K: 22.5 GB

On Hugging Face, 13 November 2026

• Library: transformers
• Parameters: 27.3B
• Weights: about 13.7 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 1,279
• Likes: 4
• License: apache-2.0
• Languages: en, zh
• Based on: xtools-at/Qwen3.8-27B-CodeForge-3xMerge
• Tags: transformers, gguf, qwen3.8, qwen3_8, merge, coding, code-generation, reasoning, efficient-thinking, agentic, tool-use, instruction-following

Links

• Model: https://huggingface.co/mradermacher/Qwen3.8-27B-CodeForge-3xMerge-i1-GGUF
• Files: https://huggingface.co/mradermacher/Qwen3.8-27B-CodeForge-3xMerge-i1-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download mradermacher/Qwen3.8-27B-CodeForge-3xMerge-i1-GGUF

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