Instructions to use nanqing233/Qwen3.8-27B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use nanqing233/Qwen3.8-27B-GGUF 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 nanqing233/Qwen3.8-27B-GGUF:Q2_K
# Run inference directly in the terminal:
llama cli -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K
# Run inference directly in the terminal:
llama cli -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

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 nanqing233/Qwen3.8-27B-GGUF:Q2_K
# Run inference directly in the terminal:
./llama-cli -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

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 nanqing233/Qwen3.8-27B-GGUF:Q2_K
# Run inference directly in the terminal:
./build/bin/llama-cli -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

Use Docker

docker model run hf.co/nanqing233/Qwen3.8-27B-GGUF:Q2_K

LM Studio
Jan
Ollama
How to use nanqing233/Qwen3.8-27B-GGUF with Ollama:
```
ollama run hf.co/nanqing233/Qwen3.8-27B-GGUF:Q2_K
```

Unsloth Studio

How to use nanqing233/Qwen3.8-27B-GGUF 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 nanqing233/Qwen3.8-27B-GGUF 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 nanqing233/Qwen3.8-27B-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for nanqing233/Qwen3.8-27B-GGUF to start chatting

How to use nanqing233/Qwen3.8-27B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "nanqing233/Qwen3.8-27B-GGUF:Q2_K"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use nanqing233/Qwen3.8-27B-GGUF with Docker Model Runner:
```
docker model run hf.co/nanqing233/Qwen3.8-27B-GGUF:Q2_K
```

Lemonade

How to use nanqing233/Qwen3.8-27B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull nanqing233/Qwen3.8-27B-GGUF:Q2_K

Run and chat with the model

lemonade run user.Qwen3.8-27B-GGUF-Q2_K

List all available models

lemonade list

Hermes Agent

How to use nanqing233/Qwen3.8-27B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

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 nanqing233/Qwen3.8-27B-GGUF:Q2_K

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use nanqing233/Qwen3.8-27B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nanqing233/Qwen3.8-27B-GGUF:Q2_K

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 "nanqing233/Qwen3.8-27B-GGUF:Q2_K" \
  --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"

Qwen3.8-27B - GGUF Quantized

This repository provides GGUF quantized versions for Qwen/Qwen3.8-27B, including the unquantized F16 baseline and the full standard K-quants series.

All models were converted and quantized directly via native llama.cpp without any dataset retraining, fine-tuning, or architectural modifications (no imatrix dependencies), ensuring maximum compatibility and weight purity.

Quantization Variants & VRAM Requirements

Filename	Quant Level	Rec. VRAM/RAM	Description & Best Use Cases
`Qwen3.8-27B-F16.gguf`	F16	~50.9 GB	Unquantized Base: Full 16-bit precision float model converted to GGUF format for reference or re-quantization.
`Qwen3.8-27B-Q8_0.gguf`	Q8_0	~27 GB	Ultra High Precision: Virtually lossless quality compared to FP16.
`Qwen3.8-27B-Q6_K.gguf`	Q6_K	~20.8 GB	High Fidelity: Imperceptible quality drop; ideal for fully offloading on 24GB GPUs.
`Qwen3.8-27B-Q5_K_M.gguf`	Q5_K_M	~18.1 GB	Balanced Pick: Excellent trade-off between speed, memory, and reasoning/coding performance.
`Qwen3.8-27B-Q5_K_S.gguf`	Q5_K_S	~17.6 GB	Slightly smaller variant of Q5_K_M for tight VRAM limits.
`Qwen3.8-27B-Q4_K_M.gguf`	Q4_K_M	~15.6 GB	[Recommended] The sweet spot for daily usage with minimal quality degradation.
`Qwen3.8-27B-Q4_K_S.gguf`	Q4_K_S	~14.7 GB	Lighter 4-bit quantization to reduce memory footprint further.
`Qwen3.8-27B-Q3_K_M.gguf`	Q3_K_M	~12.5 GB	3-bit medium quantization for low-resource or edge deployment.
`Qwen3.8-27B-Q2_K.gguf`	Q2_K	~10.1 GB	Extreme compression; recommended only for testing under severe memory constraints.

Quick Start & Usage

1. LM Studio / KoboldCPP / Ollama

LM Studio: Search for or drag-and-drop the chosen .gguf file to load.

Ollama: Create a Modelfile:

FROM ./Qwen3.8-27B-Q4_K_M.gguf

Then build and run:

ollama create Qwen3.8-27B -f Modelfile
ollama run Qwen3.8-27B

2. llama.cpp CLI

./llama-cli -m Qwen3.8-27B-Q4_K_M.gguf -p "Hello! Introduce yourself briefly." -n 512 -ngl 99

Quantization Details

Base Model: Qwen/Qwen3.8-27B (BF16 / FP16)
Conversion Tool: llama.cpp
Quant Scheme: Standard K-quants (F16, Q8_0, Q6_K, Q5_K_S/M, Q4_K_S/M, Q3_K_M, Q2_K)
Matrix Calibration: None (No imatrix required, ensuring universal applicability)

License

This repository inherits the license from the base model Qwen/Qwen3.8-27B under the Apache 2.0 license.

Downloads last month: 11

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

2-bit

Inference Providers NEW

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

Model tree for nanqing233/Qwen3.8-27B-GGUF

Base model

Qwen/Qwen3.8-27B

Quantized

(855)

this model