Instructions to use Qwen/Qwen3-Coder-Next with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Qwen/Qwen3-Coder-Next with Transformers:

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

pipe = pipeline("text-generation", model="Qwen/Qwen3-Coder-Next")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-Coder-Next")
model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-Coder-Next")
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]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Qwen/Qwen3-Coder-Next with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Qwen/Qwen3-Coder-Next

SGLang

How to use Qwen/Qwen3-Coder-Next 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 "Qwen/Qwen3-Coder-Next" \
    --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": "Qwen/Qwen3-Coder-Next",
		"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 "Qwen/Qwen3-Coder-Next" \
        --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": "Qwen/Qwen3-Coder-Next",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Qwen/Qwen3-Coder-Next with Docker Model Runner:
```
docker model run hf.co/Qwen/Qwen3-Coder-Next
```

Very specific json formatting issue in tool calls

#14

by deleted - opened Feb 5

Discussion

deleted

Feb 5

With opencode (at least), this model exhibits a very frequent and specific issue. It outputs:

"filePath"/home/username

rather than:

"filePath": "/home/username

thereby breaking the json syntax and the tool call. I've seen this in two different Q6 GGUF variants with llama.cpp, and also in a AWQ 4-bit variant with vLLM, so I think it's in the model itself, though it could be very repeatable quantization artifact.

nephepritou

Feb 5

https://huggingface.co/unsloth/Qwen3-Coder-Next-FP8-Dynamic/discussions/2#69843147f0b08d0d4d52ae03 looks similar

lightenup

Feb 6

If @JoeSmith245 is using the provided jinja chat template and the recommended tool calling parser the model wouldn't directly generate json output.

From the jinja chat template:

If you choose to call a function ONLY reply in the following format with NO suffix:

<tool_call>
<function=example_function_name>
<parameter=example_parameter_1>
value_1
</parameter>
<parameter=example_parameter_2>
This is the value for the second parameter
that can span
multiple lines
</parameter>
</function>
</tool_call>

"filePath"/home/username has to come imo from the tool parser. llama.cpp bug? I briefly checked the relevant source and it seems that llama.cpp already does grammar-based sampling of tool calls (ofc requires --jinjain the llama-server launch command). The model imo has no other choice than to produce a syntactically correct tool call, so I am really not sure how the posted output is possible.

robert1968

Feb 16

•

edited Feb 20

Workaround for "nvalid [tool=write, error=Invalid input for tool write: JSON parsing failed: T..." issues:

https://github.com/pwilkin/llama.cpp/tree/autoparser
Thank you very much

LadislavDanis

Feb 20

Hello, I was very frustrated that calling tools wasn't working, until I solved it like this
https://github.com/ladislav-danis/systemd-llm-switch

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment