Instructions to use techwithsergiu/Qwen3.5-text-9B-bnb-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use techwithsergiu/Qwen3.5-text-9B-bnb-4bit with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="techwithsergiu/Qwen3.5-text-9B-bnb-4bit")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("techwithsergiu/Qwen3.5-text-9B-bnb-4bit")
model = AutoModelForCausalLM.from_pretrained("techwithsergiu/Qwen3.5-text-9B-bnb-4bit", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use techwithsergiu/Qwen3.5-text-9B-bnb-4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "techwithsergiu/Qwen3.5-text-9B-bnb-4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "techwithsergiu/Qwen3.5-text-9B-bnb-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/techwithsergiu/Qwen3.5-text-9B-bnb-4bit

SGLang

How to use techwithsergiu/Qwen3.5-text-9B-bnb-4bit with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "techwithsergiu/Qwen3.5-text-9B-bnb-4bit" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "techwithsergiu/Qwen3.5-text-9B-bnb-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "techwithsergiu/Qwen3.5-text-9B-bnb-4bit" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "techwithsergiu/Qwen3.5-text-9B-bnb-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use techwithsergiu/Qwen3.5-text-9B-bnb-4bit with Docker Model Runner:
```
docker model run hf.co/techwithsergiu/Qwen3.5-text-9B-bnb-4bit
```

Qwen3.5-text-9B-bnb-4bit / README.md

techwithsergiu

upd visual tower info in readme

a48c4bf verified 6 months ago

4.32 kB

tags:
  - techwithsergiu
library_name: transformers
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3.5-9B/blob/main/LICENSE
pipeline_tag: text-generation
base_model:
  - techwithsergiu/Qwen3.5-text-9B

Qwen3.5-text-9B-bnb-4bit

BNB NF4 4-bit quantization of techwithsergiu/Qwen3.5-text-9B — a text-only derivative of Qwen/Qwen3.5-9B.

No visual tower — text input only. This is the recommended base for Unsloth LoRA text fine-tuning: smaller VRAM footprint, no visual-dependency complexity, cleaner adapter targeting.

Inference has been verified. LoRA fine-tuning docs are pending — see Fine-tuning section below.

What was changed from the original Qwen3.5-9B

Visual tower removed (same as Qwen3.5-text-9B)
Text backbone quantized to BNB NF4 double-quant (bnb_4bit_quant_type=nf4, bnb_4bit_compute_dtype=bfloat16)
lm_head.weight kept at bf16 for output quality / stability

Model family

Model	Type	Base model
Qwen/Qwen3.5-9B	f16 · VLM · source	—
techwithsergiu/Qwen3.5-9B-bnb-4bit	BNB NF4 · VLM	Qwen/Qwen3.5-9B
techwithsergiu/Qwen3.5-text-9B	bf16 · text-only	Qwen/Qwen3.5-9B
techwithsergiu/Qwen3.5-text-9B-bnb-4bit	BNB NF4 · text-only	Qwen3.5-text-9B
techwithsergiu/Qwen3.5-text-9B-GGUF	GGUF quants	Qwen3.5-text-9B

The visual tower scales with model size (~0.19 GB for 0.8B, ~0.62 GB for 2B/4B, ~0.85 GB for 9B). BNB text-only models are roughly 34% of the original f16 size (4B example: 9.32 GB → 3.12 GB).

Inference

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "techwithsergiu/Qwen3.5-text-9B-bnb-4bit"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",
    trust_remote_code=True,
)

messages = [{"role": "user", "content": "What is the capital of Romania?"}]

# Thinking OFF — direct answer
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=False,
)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256)
response = tokenizer.decode(
    outputs[0][inputs["input_ids"].shape[1]:],
    skip_special_tokens=True,
)
print(response)

# Thinking ON — chain-of-thought before the answer
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True,
)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=1024)
response = tokenizer.decode(
    outputs[0][inputs["input_ids"].shape[1]:],
    skip_special_tokens=True,
)
print(response)

Fine-tuning

TBD — LoRA training with this model has not been documented yet. The model has been verified for inference (text generation, thinking ON/OFF). The expectation is that standard Unsloth LoRA training applies — this is a text-only BNB 4-bit model architecturally identical to models Unsloth supports — but this has not been tested yet and there is no official Qwen3.5 text-only training guide to reference.

For VLM (image + text) fine-tuning of the full model, see: unsloth.ai/docs/models/qwen3.5/fine-tune

Pipeline diagram

Acknowledgements

Based on Qwen/Qwen3.5-9B by the Qwen Team. If you use this model in research, please cite the original:

@misc{qwen3.5,
    title  = {{Qwen3.5}: Towards Native Multimodal Agents},
    author = {{Qwen Team}},
    month  = {February},
    year   = {2026},
    url    = {https://qwen.ai/blog?id=qwen3.5}
}