Instructions to use mduca771/Qwen3.6-27B-imatrix-pt with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use mduca771/Qwen3.6-27B-imatrix-pt with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mduca771/Qwen3.6-27B-imatrix-pt: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 mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf mduca771/Qwen3.6-27B-imatrix-pt: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 mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

Use Docker

docker model run hf.co/mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

LM Studio
Jan
Ollama
How to use mduca771/Qwen3.6-27B-imatrix-pt with Ollama:
```
ollama run hf.co/mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
```

Unsloth Studio

How to use mduca771/Qwen3.6-27B-imatrix-pt 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 mduca771/Qwen3.6-27B-imatrix-pt 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 mduca771/Qwen3.6-27B-imatrix-pt to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for mduca771/Qwen3.6-27B-imatrix-pt to start chatting

How to use mduca771/Qwen3.6-27B-imatrix-pt with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mduca771/Qwen3.6-27B-imatrix-pt: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": "mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use mduca771/Qwen3.6-27B-imatrix-pt with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Docker Model Runner
How to use mduca771/Qwen3.6-27B-imatrix-pt with Docker Model Runner:
```
docker model run hf.co/mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M
```

Lemonade

How to use mduca771/Qwen3.6-27B-imatrix-pt with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.6-27B-imatrix-pt-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use mduca771/Qwen3.6-27B-imatrix-pt with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf mduca771/Qwen3.6-27B-imatrix-pt: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 mduca771/Qwen3.6-27B-imatrix-pt:Q4_K_M

Run Hermes

hermes

Atomic Chat

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Qwen3.6-27B Imatrix Quantizations (Portuguese-Optimized)

Quantizações com imatrix (importance matrix) do modelo Qwen3.6-27B, otimizadas para língua portuguesa.

Sobre as Quantizações

Estes quants foram gerados usando o workflow de imatrix do llama.cpp, que calcula uma matriz de importância dos pesos do modelo antes de quantizar, preservando melhor a qualidade em formatos compactos.

Base

Modelo original: Qwen3.6-27B (BF16)
Ferramenta: llama.cpp — llama-quantize --imatrix

Dado de Treino do Imatrix

A matriz de importância foi calculada usando texto em língua portuguesa, incluindo:

Textos em português: literatura, artigos técnicos, documentos acadêmicos, conversas do dia a dia
Código Python: scripts, notebooks, documentação de bibliotecas
Conteúdo misto: prompts de IA, tutoriais, materiais didáticos

Isso significa que estas quantizações têm desempenho superior para tarefas em português e para geração de código Python, comparado a quants genéricos feitos com dados em inglês.

Arquivos

Arquivo	Tamanho Aprox.	Tipo de Quantização
`Qwen3.6-27B-BF16-imatrix-pt-iq4_nl.gguf`	~14.9 GB	IQ4_NL (Normalized 4-bit)
`Qwen3.6-27B-BF16-imatrix-pt-q4_k_s.gguf`	~14.7 GB	Q4_K_S (4-bit K-small)
`Qwen3.6-27B-BF16-imatrix-pt-q4_k_m.gguf`	~15.6 GB	Q4_K_M (4-bit K-medium)
`Qwen3.6-27B-BF16-imatrix-pt-q5_k_s.gguf`	~17.6 GB	Q5_K_S (5-bit K-small)
`Qwen3.6-27B-BF16-imatrix-pt-q5_k_m.gguf`	~18.2 GB	Q5_K_M (5-bit K-medium)

Recomendações de Uso

IQ4_NL: melhor relação qualidade/tamanho para uso geral
Q4_K_M: equilíbrio entre qualidade e memória (recomendado para GPUs com 16-24 GB VRAM)
Q5_K_M: máxima qualidade entre os quants disponíveis (recomendado para GPUs com 24+ GB VRAM)

Como Usar

Ollama

ollama run ./Qwen3.6-27B-BF16-imatrix-pt-q4_k_m.gguf

llama.cpp

./llama-cli -m Qwen3.6-27B-BF16-imatrix-pt-q4_k_m.gguf -n 512 --prompt "Olá, como vai?"

llama-server

./llama-server -m Qwen3.6-27B-BF16-imatrix-pt-q4_k_m.gguf --port 8080

Notas

Sufixo pt indica que a imatrix foi treinada com dados em português
Sufixo imatrix indica quantização guiada por matriz de importância
Para uso em inglês, considere os quants padrão sem imatrix PT
Compatível com qualquer ferramenta que suporte GGUF (Ollama, llama.cpp, LM Studio, KoboldCPP, etc.)

Licença

O modelo Qwen3.6-27B segue a licença original do Qwen. As quantizações são derivadas e mantêm a mesma licença.

Downloads last month: -

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

4-bit

5-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support