Instructions to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF",
	filename="Qwen3.6-27B-MTP-UD-IQ3_XXS.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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
# Run inference directly in the terminal:
llama cli -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
# Run inference directly in the terminal:
llama cli -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
# Run inference directly in the terminal:
./llama-cli -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
# Run inference directly in the terminal:
./build/bin/llama-cli -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

Use Docker

docker model run hf.co/GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

LM Studio
Jan

vLLM

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-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": "GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

Ollama
How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with Ollama:
```
ollama run hf.co/GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
```

Unsloth Studio

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF to start chatting

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

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": "GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

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 GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with Docker Model Runner:
```
docker model run hf.co/GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS
```

Lemonade

How to use GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull GazTrab/Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF:UD-IQ3_XXS

Run and chat with the model

lemonade run user.Qwen3.6-27B-MTP-UD-IQ3_XXS-GGUF-UD-IQ3_XXS

List all available models

lemonade list

Qwen3.6-27B-MTP-UD-IQ3_XXS GGUF

Qwen3.6-27B dense model with Multi-Token Prediction (MTP) head, quantized to IQ3_XXS using Unsloth Dynamic quantization.

This GGUF was created by grafting an MTP prediction head (block 64) onto the Unsloth IQ3_XXS base model, enabling speculative decoding without a separate draft model.

Key Specs

Property	Value
Architecture	Qwen3.6 (hybrid SSM + attention, dense)
Parameters	27.3B
Active parameters	27.3B (dense, all active per token)
Quantization	IQ3_XXS (Unsloth Dynamic 2.0)
MTP layers	1 (nextn_predict_layers=1)
Block count	65 (64 base + 1 MTP)
Tensors	866
File size	12.45 GB
Context length	262,144 native (tested up to 56k with MTP)

Performance (RTX 5080 16GB)

Benchmarked with llama.cpp (s015-mtp build, am17an PR #22673):

Metric	Value
Token generation (avg)	76 tok/s with MTP, 53 tok/s without
Token generation (peak)	102 tok/s on code tasks
MTP acceptance rate	90.6% aggregate (95-100% on code, 62-85% on creative)
GPU layers	66/66 (fits entirely on 16 GB)
VRAM usage	~12.5 GB model + 150 MiB recurrent state
GSM8K accuracy	89/100 (89.0%, Wilson CI [85.7%, 96.4%])

CodeNeedle Positional Recall (http_server.py, 11 functions, ~50k char context)

Config	Pass	Lines matched	Hallucinated
This model (q8_0 KV, 32k ctx)	11/11	220/220 (100%)	0
This model (q4_0 KV, 56k ctx)	11/11	218/220 (99.1%)	1
35B-A3B MoE UD-Q4_K_XL	11/11	206/220 (93.6%)	12
27B MTP Q2_K_XL	10/11	199/220 (90.5%)	20

How to Use

Requires a llama.cpp build with MTP support (am17an's mtp-clean branch, PR #22673).

llama-server \
  -m Qwen3.6-27B-MTP-UD-IQ3_XXS.gguf \
  -c 32768 \
  --fit on \
  --spec-type mtp \
  -fa on \
  -t 20 \
  --no-mmap \
  --jinja \
  -ctk q8_0 -ctv q8_0

Important: --spec-type mtp must be explicitly passed to enable MTP speculation. Without it, the model loads the MTP head but doesn't draft tokens (~56 tok/s instead of ~76 tok/s).

Extended Context with q4_0 KV

For longer contexts (up to 56k stable), use q4_0 KV cache:

-c 57344 -ctk q4_0 -ctv q4_0 --spec-type mtp

q4_0 KV is near-lossless (218/220 CodeNeedle at 56k) and extends max stable context from 32k to 56k. Beyond 56k, the MTP compute buffer OOMs on 16 GB VRAM.

How This Was Made

Base model: unsloth/Qwen3.6-27B-UD-IQ3_XXS (12 GB, 851 tensors, 64 blocks)
MTP head: Extracted from havenoammo's MTP GGUF collection — 15 tensors for block 64 (attention + FFN + nextn prediction head), Q8_0 quantized, 436 MB
Graft: Custom Python script using the gguf library (GGUFReader + GGUFWriter). Copies all base tensors + MTP tensors, sets block_count=65, adds nextn_predict_layers=1. SHA256-verified integrity.

The graft script is available at: scripts/graft-mtp.py (adapt for other base models)

Why IQ3_XXS + MTP?

The "dream config" thesis: a 27B dense model at IQ3_XXS fits entirely on a 16 GB GPU (no PCIe bottleneck), while MTP provides free speculative decoding. This combination delivers:

Higher quality than MoE: 220/220 CodeNeedle vs 206/220 for 35B MoE (no expert routing = more coherent at low quant)
Faster than MoE: 76 tok/s vs 50 tok/s (no expert loading over PCIe)
Smaller than MoE: 12.45 GB vs 21 GB (fits fully on GPU with room to spare)

Limitations

MTP requires a custom llama.cpp build (not yet in mainline as of May 2026, but PR #22673 is close to merging)
TurboQuant KV cache (turbo4) is not compatible with the current MTP builds (build incompatibility, not a model issue)
Max stable context with MTP is ~56k on 16 GB VRAM (compute buffer OOM beyond that)
Without MTP, this is just a standard IQ3_XXS model running at ~53 tok/s

Credits

Unsloth — IQ3_XXS base quantization (Dynamic 2.0)
havenoammo — MTP head tensors + graft concept
am17an — llama.cpp MTP implementation (PR #22673)
Qwen Team — Qwen3.6-27B base model

Downloads last month: 428

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

3-bit