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

Libraries

How to use steampunque/Qwen3.5-27B-MP-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="steampunque/Qwen3.5-27B-MP-GGUF",
	filename="Qwen3.5-27B.Q4_K_H.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use steampunque/Qwen3.5-27B-MP-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
# Run inference directly in the terminal:
llama-cli -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
# Run inference directly in the terminal:
llama-cli -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

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 steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
# Run inference directly in the terminal:
./llama-cli -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

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 steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
# Run inference directly in the terminal:
./build/bin/llama-cli -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

Use Docker

docker model run hf.co/steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

LM Studio
Jan
Ollama
How to use steampunque/Qwen3.5-27B-MP-GGUF with Ollama:
```
ollama run hf.co/steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
```

Unsloth Studio

How to use steampunque/Qwen3.5-27B-MP-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 steampunque/Qwen3.5-27B-MP-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 steampunque/Qwen3.5-27B-MP-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

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

How to use steampunque/Qwen3.5-27B-MP-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

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": "steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use steampunque/Qwen3.5-27B-MP-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 steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

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 steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use steampunque/Qwen3.5-27B-MP-GGUF with Docker Model Runner:
```
docker model run hf.co/steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H
```

Lemonade

How to use steampunque/Qwen3.5-27B-MP-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull steampunque/Qwen3.5-27B-MP-GGUF:Q6_K_H

Run and chat with the model

lemonade run user.Qwen3.5-27B-MP-GGUF-Q6_K_H

List all available models

lemonade list

Mixed Precision GGUF layer quantization of Qwen3.5-27B by Qwen

Original model: https://huggingface.co/Qwen/Qwen3.5-27B

The hybrid quant employs different quantization levels on a per layer basis to enable both high performance and small file size at the same time. The quants employed are all K to avoid slow CPU or older GPU processing of IQ quants. For this file the layer quants are as follows:

Q4_K_L : Q4_K_M + attn_o = q6_k
Q5_K_L : attn_v = q8_0 attn_o = q6_k ffn_d = q6_k
Q6_K_S : Q6_K

   LAYER_TYPES='[
   [0 ,"Q5_K_L"],[1 ,"Q5_K_M"],[2 ,"Q5_K_S"],[3 ,"Q4_K_L"],[4 ,"Q4_K_M"],[5 ,"Q4_K_S"],[6 ,"Q3_K_L"],[7 ,"Q3_K_M"],
   [8 ,"Q3_K_L"],[9 ,"Q3_K_M"],[10,"Q3_K_L"],[11,"Q3_K_M"],[12,"Q3_K_L"],[13,"Q3_K_M"],[14,"Q3_K_L"],[15,"Q3_K_M"],
   [16,"Q3_K_L"],[17,"Q3_K_L"],[18,"Q3_K_L"],[19,"Q3_K_L"],[20,"Q3_K_L"],[21,"Q3_K_L"],[22,"Q3_K_L"],[23,"Q3_K_L"],
   [24,"Q4_K_S"],[25,"Q3_K_L"],[26,"Q4_K_S"],[27,"Q3_K_L"],[28,"Q4_K_S"],[29,"Q3_K_L"],[30,"Q4_K_S"],[31,"Q3_K_L"],
   [32,"Q4_K_S"],[33,"Q4_K_S"],[34,"Q4_K_S"],[35,"Q4_K_S"],[36,"Q4_K_S"],[37,"Q4_K_S"],[38,"Q4_K_S"],[39,"Q4_K_S"],
   [40,"Q4_K_M"],[41,"Q4_K_S"],[42,"Q4_K_M"],[43,"Q4_K_S"],[44,"Q4_K_M"],[45,"Q4_K_S"],[46,"Q4_K_M"],[47,"Q4_K_S"],
   [48,"Q4_K_M"],[49,"Q4_K_M"],[50,"Q4_K_M"],[51,"Q4_K_M"],[52,"Q4_K_M"],[53,"Q4_K_M"],[54,"Q4_K_M"],[55,"Q4_K_M"],
   [56,"Q4_K_M"],[57,"Q4_K_M"],[58,"Q4_K_M"],[59,"Q4_K_L"],[60,"Q5_K_S"],[61,"Q5_K_M"],[62,"Q5_K_L"],[63,"Q6_K_S"]
   ]'
   FLAGS="--token-embedding-type Q4_K --output-tensor-type Q6_K --layer-types-high"

The quant was tested for very strong performance over a small set of curated reasoning prompts.

A second larger Q6_K_H quant is also available. This quant was sized to maintain usable context with 24G VRAM with minimum quant of Q4_K_S across layers and strong Q6_K_L output layers. It was iterated over a curated set of test prompts and gave excellent performance across the whole set, missing only 1 IQ test prompt which most models dont get.

Q4_K_L : Q4_K_M + attn_o = q6_k
Q5_K_L : attn_v = q8_0 attn_o = q6_k ffn_d = q6_k
Q6_K_S : Q6_K
Q6_K_M : attn_v = q8_0 ffn_d = q8_0
Q6_K_L : attn_v = q8_0 attn_o = q8_0 ffn_d = q8_0

   LAYER_TYPES='[
   [0 ,"Q6_K_S"],[1 ,"Q5_K_M"],[2 ,"Q4_K_M"],[3 ,"Q4_K_S"],[4 ,"Q4_K_M"],[5 ,"Q4_K_S"],[6 ,"Q4_K_M"],[7 ,"Q4_K_S"],
   [8 ,"Q4_K_M"],[9 ,"Q4_K_S"],[10,"Q4_K_M"],[11,"Q4_K_S"],[12,"Q4_K_M"],[13,"Q4_K_S"],[14,"Q4_K_M"],[15,"Q4_K_S"],
   [16,"Q4_K_M"],[17,"Q4_K_M"],[18,"Q4_K_M"],[19,"Q4_K_M"],[20,"Q4_K_M"],[21,"Q4_K_M"],[22,"Q4_K_M"],[23,"Q4_K_M"],
   [24,"Q4_K_M"],[25,"Q4_K_M"],[26,"Q4_K_M"],[27,"Q4_K_M"],[28,"Q4_K_M"],[29,"Q4_K_M"],[30,"Q4_K_M"],[31,"Q4_K_M"],
   [32,"Q4_K_L"],[33,"Q4_K_M"],[34,"Q4_K_L"],[35,"Q4_K_M"],[36,"Q4_K_L"],[37,"Q4_K_M"],[38,"Q4_K_L"],[39,"Q4_K_M"],
   [40,"Q4_K_L"],[41,"Q4_K_L"],[42,"Q4_K_L"],[43,"Q4_K_L"],[44,"Q5_K_S"],[45,"Q5_K_S"],[46,"Q5_K_S"],[47,"Q5_K_S"],
   [48,"Q5_K_M"],[49,"Q5_K_M"],[50,"Q5_K_M"],[51,"Q5_K_M"],[52,"Q5_K_L"],[53,"Q5_K_L"],[54,"Q5_K_L"],[55,"Q5_K_L"],
   [56,"Q6_K_S"],[57,"Q6_K_S"],[58,"Q6_K_S"],[59,"Q6_K_S"],[60,"Q6_K_M"],[61,"Q6_K_M"],[62,"Q6_K_L"],[63,"Q6_K_L"]
   ]'
   FLAGS="--token-embedding-type Q6_K --output-tensor-type Q6_K --layer-types-high"

Comparison:

Quant	size	PPL	Comment
IQ4_XS	14.8e9	8.5	IQ4_XS with default embedding and output
Q4_K_H	16.0e9	7.4	Hybrid quant with Q4_K embedding Q6_K output
Q6_K	22.1e9	8.4	Q6_K with default embedding and output
Q6_K_H	18.6e9	7.7	Hybrid quant with Q6_K embedding Q6_K output

Usage:

Qwen3.5-27B is a vision capable dense RL model. It can be used together with its multimedia projector layers to process images and text inputs and generate text outputs. The mmproj file is made available in this repository.

Update 3/18/26: original mmproj had BF16 mmproj tensors. These are still available, unmodified, renamed to *.mmproj.BF16.gguf. New F16 mmproj format is the default to enable working across the widest range of platforms.

Speculation does not work with the model due to the attention sheme it uses. On a 2x 4070 setup (1 RPC) approx performance is:

Q	QKV	NKV	gen tps
Q4_K_H	F16	100k +	22
Q4_K_H	Q8_0	180k +	22
Q6_K_H	F16	60k	20
Q6_K_H	Q8_0	100k	20

Long context test crashed in graph on llama.cpp when using RPC most likely due to bug discussed below, but before it crashed prompt processing seemed quite fast. Model will not be usable/reliable until this problem is resolved.

The model appears to be trained to decide itself whether to do a think block or not. When it does a think block it falls into very heavy overthinking but does come up with accurate answers. Over a small set of eval prompts the model did extremely well. To avoid the overthinking inject think start and think stop tokens first thing after assistant prompt:

THINK_START="<think>\n"
THINK_STOP="\n</think>\n\n"

If the model doesnt feel like doing thinking on a given prompt it will automatically do this. To force the model into a think block inject a bootstrap think block following the assistant prompt:

"<think>\nHere's a thinking process to solve the problem:"

The model was found to be highly capable on reasoning tasks when skipping think block, with zero overthinking, just accurate direct deductions to final solutions.

The model was tested in vision mode on a couple pretty tough bird ID image and did spectacularly well, with a very detailed think block unlike any model I have seen to date.

The Q4_K_H model was tested across a small set of code gen prompts and found to be quite intermittent in its ability to generate working code. The Q6_K_H model appears to do a much better job generating working code on code prompts both with and without think block enabled.

Llama.cpp minimum version to run Qwen3.5-27B should be b8148 and above due to correction of a graph error which causes crashes in both RPC and multiple local GPU setups. If the model run over RPC it will crash due to an unresolved memory leak in RPC: https://github.com/ggml-org/llama.cpp/issues/19892, temp workaround set GGML_CUDA_DISABLE_GRAPHS=1 on rpc server launch.

Benchmarks:

A full set of both math and vision benchmarks for the model will eventually be given here: https://huggingface.co/spaces/steampunque/benchlm

Download the file from below:

Link	Type	Size/e9 B	Notes
Qwen3.5-27B.Q4_K_H.gguf	Q4_K_H	16e9 B	1G bigger than IQ4_XS
Qwen3.5-27B.Q6_K_H.gguf	Q6_K_H	18.6e9 B	3.5G smaller than Q6_K
Qwen3.5-27B.mmproj.gguf	F16	0.93e9 B	multimedia projector
Qwen3.5-27B.mmproj.BF16.gguf	BF16	0.93e9 B	multimedia projector

A discussion thread about the hybrid layer quant approach can be found here on the llama.cpp git repository:

https://github.com/ggml-org/llama.cpp/discussions/13040

Downloads last month: 110

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

6-bit

Inference Providers NEW

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

Model tree for steampunque/Qwen3.5-27B-MP-GGUF

Base model

Qwen/Qwen3.5-27B

Quantized

(205)

this model