Instructions to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="sh111111111111111/Qwen3.5-9B-BitClass3-GGUF",
	filename="Qwen3.5-9B-Q3_K_S.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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M

Use Docker

docker model run hf.co/sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sh111111111111111/Qwen3.5-9B-BitClass3-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": "sh111111111111111/Qwen3.5-9B-BitClass3-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M

Ollama
How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with Ollama:
```
ollama run hf.co/sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
```

Unsloth Studio

How to use sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for sh111111111111111/Qwen3.5-9B-BitClass3-GGUF to start chatting

How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf sh111111111111111/Qwen3.5-9B-BitClass3-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": "sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-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 sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with Docker Model Runner:
```
docker model run hf.co/sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M
```

Lemonade

How to use sh111111111111111/Qwen3.5-9B-BitClass3-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull sh111111111111111/Qwen3.5-9B-BitClass3-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-9B-BitClass3-GGUF-Q4_K_M

List all available models

lemonade list

Qwen3.5-9B — BitClass3 Mixed-Precision GGUF

Mixed-precision GGUF quantizations of Qwen3.5-9B. BitClass3 keeps the Hessian-sensitivity front-end to set each level's bit budget, but hands the per-tensor allocation to an error-minimizing solver (built on llama.cpp's --target-bpw) that distributes bits across every tensor — including the hybrid DeltaNet/SSM tensors — to minimize imatrix-weighted quantization error at the target size.

Available Quantizations

PPL is the in-house eval (continuity with prior releases); KLD vs the BF16 source is the primary quality metric (mean and the robust 99.9th percentile).

File	BPW	Size	PPL ↓	KL-mean ↓	KL-99.9% ↓	Use Case
`Qwen3.5-9B-Q8_0.gguf`	8.5	9.53 GB	1.73	0.0015	0.065	Near-lossless reference
`Qwen3.5-9B-Q6_K.gguf`	6.0	6.71 GB	1.73	0.0238	1.166	High quality
`Qwen3.5-9B-Q5_K_M.gguf`	5.4	6.00 GB	1.73	0.0268	1.238	Balanced quality and size
`Qwen3.5-9B-Q4_K_M.gguf`	4.8	5.41 GB	1.74	0.0346	1.312	Best quality-to-size ratio
`Qwen3.5-9B-Q3_K_S.gguf`	3.6	4.07 GB	1.78	0.0893	2.648	Maximum compression

Recommended: Q4_K_M — KL-mean 0.035 at 5.41 GB; PPL is within rounding of Q8_0.

How It Compares

Model	BPW	Size	PPL ↓	Source
ByteShape IQ3_S 3.00bpw	3.0	3.37 GB	2.069	byteshape
★ Ours Q3_K_S	3.6	4.07 GB	1.78	This repo
★ Ours Q4_K_M	4.8	5.41 GB	1.74	This repo
★ Ours Q5_K_M	5.4	6.00 GB	1.73	This repo

Our Q3_K_S beats ByteShape's 3.00bpw 9B on perplexity (1.78 vs 2.069). ByteShape's higher-BPW rows reach lower PPL. The PPL curve is near-flat from Q8_0 down to Q4_K_M (within rounding), and KL-mean stays at/under 0.035 through Q4_K_M — the error-minimizing allocation spending bits where they reduce divergence most, rather than by a fixed per-suffix rule.

Key Sensitivity Findings (Qwen3.5-9B)

The Hessian sensitivity pattern for 9B is fundamentally different from 4B:

blk.3 (early layer) is most sensitive — score 1.0 for k/v. On 4B it was blk.34 (late layer).
Sensitivity peaks at both ends AND middle: blk.3 (1.0), blk.7 (0.78), blk.23 (0.78), blk.27 (0.86), blk.31 (0.87)
ffn_down at blk.4-5 is near-zero sensitivity (0.0003) — safe for aggressive quantization
This confirms: model-specific Hessian data matters. You cannot assume late layers are always most sensitive.

How It Works

Hessian sensitivity — compute H_diag = mean(X²) per layer on calibration data; this sets each level's overall bit budget.
Error-minimizing per-tensor allocation — an imatrix-weighted solver (llama.cpp --target-bpw) assigns a quant type to every tensor to minimize total quantization error at the target BPW, covering attention, FFN, and the hybrid DeltaNet/SSM tensors.
imatrix — importance matrix computed over wikitext guides the per-tensor error.
GGUF export — produced with stock llama-quantize.

Usage

hf download sh111111111111111/Qwen3.5-9B-BitClass3-GGUF \
    Qwen3.5-9B-Q4_K_M.gguf --local-dir .

llama-cli   -m Qwen3.5-9B-Q4_K_M.gguf -cnv
llama-server -m Qwen3.5-9B-Q4_K_M.gguf --port 8080

Note: Qwen3.5 GGUFs are not currently runnable in Ollama (vision/mmproj handling is not yet supported there); use llama.cpp or LM Studio.

Benchmark Details

NVIDIA GB10 ATOM (128 GB unified memory, aarch64). llama.cpp with --target-bpw (PR #15550). PPL via llama-perplexity (in-house eval). KLD via llama-perplexity --kl-divergence against BF16-source logits (mean / median / 99.9th percentile reported; the single-token KL-max is omitted as an unstable order statistic). wikitext-2 PPL also tracked internally.

Disclaimer

Independent project. Not affiliated with or endorsed by Qwen, Unsloth, ByteShape, Bartowski, or llama.cpp. Competitor figures are from our own benchmark harness and may differ from those projects' self-reported numbers; competitor file sizes reflect the revision we tested and may since have changed.

License

Apache 2.0, inherited from Qwen3.5-9B.

Downloads last month: -

GGUF

Model size

9B params

Architecture

qwen35

Hardware compatibility

3-bit

4-bit

5-bit

6-bit

8-bit

Model tree for sh111111111111111/Qwen3.5-9B-BitClass3-GGUF

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Quantized

(296)

this model