Instructions to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Aikimi/Qwen3.5-27B-UHCSA-Japanese",
	filename="Qwen3.5-27B-HauhauAggro-Q4_K_M-ja_imatrix.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Use Docker

docker model run hf.co/Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

LM Studio
Jan

vLLM

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Aikimi/Qwen3.5-27B-UHCSA-Japanese"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Aikimi/Qwen3.5-27B-UHCSA-Japanese",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Ollama
How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Ollama:
```
ollama run hf.co/Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
```

Unsloth Studio

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Aikimi/Qwen3.5-27B-UHCSA-Japanese to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Aikimi/Qwen3.5-27B-UHCSA-Japanese to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Aikimi/Qwen3.5-27B-UHCSA-Japanese to start chatting

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Docker Model Runner:
```
docker model run hf.co/Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M
```

Lemonade

How to use Aikimi/Qwen3.5-27B-UHCSA-Japanese with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Aikimi/Qwen3.5-27B-UHCSA-Japanese:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-27B-UHCSA-Japanese-Q4_K_M

List all available models

lemonade list

Qwen3.5-27B-UHCSA-Japanese

This repository provides four GGUF quantizations of Qwen3.5-27B-HauhauAggro for Japanese-focused local inference.

Source

The source model before the quantization variants in this repository is HauhauCS/Qwen3.5-27B-Uncensored-HauhauCS-Aggressive.

Files

File	Quantization	Notes
`Qwen3.5-27B-HauhauAggro-Q4_K_M-ja_imatrix.gguf`	Q4_K_M	`imatrix` variant with lower memory use than Q5/Q6
`Qwen3.5-27B-HauhauAggro-Q4_K_M-ja_plain.gguf`	Q4_K_M	`plain` variant in the same size class as Q4 imatrix
`Qwen3.5-27B-HauhauAggro-Q5_K_M-ja_imatrix.gguf`	Q5_K_M	Higher quality than Q4, with higher memory use
`Qwen3.5-27B-HauhauAggro-Q6_K-ja_imatrix.gguf`	Q6_K	Highest quality in this set, with the largest memory use

Variants

imatrix: quantized with imatrix calibration.
plain: quantized without imatrix calibration.

The imatrix variants in this repository were quantized using the TFMC/imatrix-dataset-for-japanese-llm dataset. The dataset page lists the license as odc-by, so attribution is provided here. Refer to the dataset page for the full license terms and attribution requirements.

Usage

llama.cpp

./llama-cli -m Qwen3.5-27B-HauhauAggro-Q4_K_M-ja_plain.gguf -p "Write a short self-introduction in Japanese."

Ollama import

Create a Modelfile like this:

FROM ./Qwen3.5-27B-HauhauAggro-Q4_K_M-ja_plain.gguf

TEMPLATE """{{- if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{- if .Prompt }}<|im_start|>user
{{ .Prompt }}<|im_end|>
<|im_start|>assistant
{{ end }}"""

PARAMETER stop "<|im_start|>"
PARAMETER stop "<|im_end|>"
PARAMETER stop "<|endoftext|>"

Then run:

ollama create qwen35-uhcsa-japanese -f Modelfile
ollama run qwen35-uhcsa-japanese

Notes

This repository contains GGUF weights only.
Choose Q4 for lower memory usage and Q5/Q6 for higher quality.
The four files are bundled in a single repository for convenience.

Downloads last month: 22

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

4-bit

5-bit

6-bit