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

Libraries

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="zlaabsi/Qwen3.6-27B-OTQ-GGUF",
	filename="Qwen3.6-27B-OTQ-DYN-Q3_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": [
				{
					"type": "text",
					"text": "Describe this image in one sentence."
				},
				{
					"type": "image_url",
					"image_url": {
						"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
					}
				}
			]
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M

Install from WinGet (Windows)

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

Use Docker

docker model run hf.co/zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zlaabsi/Qwen3.6-27B-OTQ-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zlaabsi/Qwen3.6-27B-OTQ-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M

Ollama
How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with Ollama:
```
ollama run hf.co/zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M
```

Unsloth Studio new

How to use zlaabsi/Qwen3.6-27B-OTQ-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 zlaabsi/Qwen3.6-27B-OTQ-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 zlaabsi/Qwen3.6-27B-OTQ-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

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

Pi new

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf zlaabsi/Qwen3.6-27B-OTQ-GGUF: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": "zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf zlaabsi/Qwen3.6-27B-OTQ-GGUF: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 zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with Docker Model Runner:
```
docker model run hf.co/zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M
```

Lemonade

How to use zlaabsi/Qwen3.6-27B-OTQ-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull zlaabsi/Qwen3.6-27B-OTQ-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.6-27B-OTQ-GGUF-Q4_K_M

List all available models

lemonade list

Qwen3.6-27B-OTQ-GGUF / BENCHMARKS.md

zlaabsi

Add paired BF16 quantization chart and reproducibility links

b057957 verified 14 days ago

preview code

raw

history blame contribute delete

3.04 kB

Benchmarks

Benchmarks are split into measured OTQ runtime frontiers, release-gate checks, allocation transparency, and the official Qwen reference table.

Measured OTQ Runtime

Vector: SVG | PDF

Release Gates

Vector: SVG | PDF

These release suites are deterministic guardrails, not substitutes for full academic benchmarks.

Official Qwen Baseline

Official Qwen3.6-27B language benchmark scores are imported as an external reference table in benchmarks/official_qwen36_baseline.csv. They are not plotted against OTQ until matching benchmark tasks are run on these GGUF files.

Deltas versus the official Qwen baseline must only be reported for tasks that are actually run on the OTQ artifacts with the same task definition and scoring rule.

OTQ Task Runs

No separate OTQ-only benchmark subset is attached beyond the paired BF16-vs-GGUF mini-subset.

Paired BF16-vs-GGUF Mini-Subset

The staged repo includes a same-task, same-prompt, deterministic 232-sample practical subset comparing the BF16 sidecar against Q3_K_M, Q4_K_M, and Q5_K_M.

Vector: SVG | PDF

Files:

benchmarks/paired_bf16_quant_summary.csv
benchmarks/paired_bf16_quant_summary.json
benchmarks/paired_bf16_quant_report.md

This is a quantization-regression signal, not a full official benchmark replacement. Do not compare its small-subset mmlu_pro or gpqa rates directly to the Qwen model-card full-harness scores.

CSV Data

benchmarks/throughput.csv
benchmarks/eval_summary.csv
benchmarks/category_pass_rate.csv
benchmarks/artifacts.csv
benchmarks/tensor_allocation.csv
benchmarks/category_tensor_allocation.csv
benchmarks/official_qwen36_baseline.csv
benchmarks/paired_bf16_quant_summary.csv
benchmarks/paired_bf16_quant_summary.json
benchmarks/paired_bf16_quant_report.md
benchmarks/quant_eval.csv when separate OTQ-only task runs are present