Instructions to use Kwaipilot/KAT-Dev-72B-Exp with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Kwaipilot/KAT-Dev-72B-Exp with Transformers:

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

pipe = pipeline("text-generation", model="Kwaipilot/KAT-Dev-72B-Exp")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Kwaipilot/KAT-Dev-72B-Exp")
model = AutoModelForCausalLM.from_pretrained("Kwaipilot/KAT-Dev-72B-Exp")
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

vLLM

How to use Kwaipilot/KAT-Dev-72B-Exp with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Kwaipilot/KAT-Dev-72B-Exp"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Kwaipilot/KAT-Dev-72B-Exp",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Kwaipilot/KAT-Dev-72B-Exp

SGLang

How to use Kwaipilot/KAT-Dev-72B-Exp 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 "Kwaipilot/KAT-Dev-72B-Exp" \
    --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": "Kwaipilot/KAT-Dev-72B-Exp",
		"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 "Kwaipilot/KAT-Dev-72B-Exp" \
        --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": "Kwaipilot/KAT-Dev-72B-Exp",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Kwaipilot/KAT-Dev-72B-Exp with Docker Model Runner:
```
docker model run hf.co/Kwaipilot/KAT-Dev-72B-Exp
```

How to make this model work with Claude Code for local deployment?

by adonishong - opened Oct 11, 2025

Discussion

adonishong

Oct 11, 2025

Thank you for you guys release this fantastic model.

I have tried Claude Code -> Claude Code Router -> VLLM config, use

"vllm serve ./ --served-model-name KAT-Dev-72B-Exp --max-num-seq 1 --t
ensor-parallel-size 4 --host 0.0.0.0 --port 8000 --gpu-memory-utilization 0.975 --enable-auto-tool-choice --tool-call-parser hermes"

then try /init in Claude code, the response are all something like
"Let me start by checking if there's already a CLAUDE.md file in the repository: <{"name": "Bash", "arguments": {"command": "ls -la", "description": "Check for existing CLAUDE.md file"}}>"

no tool call triggered with in Claude Code cli

Looking solution for this use case ... appreciate for any kind of solution

shunxing1234

Kwaipilot org Oct 11, 2025

Hi! KAT-Dev-72B-Exp was trained exclusively with the SWE-agent scaffold. It uses the XMLFunctionCallingParser(https://swe-agent.com/latest/reference/parsers/#sweagent.tools.parsing.XMLFunctionCallingParser)
as its tool-calling format.

As a result, this model does not support the Claude Code CLI.

If you’d like to use our model with Claude Code, please try the Kat-Coder model instead — its API is currently available for free at https://www.streamlake.ai/document/DOC/mg6k6nlp8j6qxicx4c9
. Kat-Coder was optimized during training for multiple scaffolds and is better suited for real-world programming scenarios.

Hope this clarifies your question!

adonishong

Oct 11, 2025

•

edited Oct 11, 2025

Hi! KAT-Dev-72B-Exp was trained exclusively with the SWE-agent scaffold. It uses the XMLFunctionCallingParser(https://swe-agent.com/latest/reference/parsers/#sweagent.tools.parsing.XMLFunctionCallingParser)
as its tool-calling format.

As a result, this model does not support the Claude Code CLI.

If you’d like to use our model with Claude Code, please try the Kat-Coder model instead — its API is currently available for free at https://www.streamlake.ai/document/DOC/mg6k6nlp8j6qxicx4c9
. Kat-Coder was optimized during training for multiple scaffolds and is better suited for real-world programming scenarios.

Hope this clarifies your question!

Appreciate for your response, it is clear now.

wawoshashi

Oct 15, 2025

It seems KAT-Code is the real solution, eagerly awaiting its open-source release. Tears streaming down my face.

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