Instructions to use moonshotai/Kimi-K2-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use moonshotai/Kimi-K2-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="moonshotai/Kimi-K2-Instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("moonshotai/Kimi-K2-Instruct", trust_remote_code=True, dtype="auto")

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use moonshotai/Kimi-K2-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "moonshotai/Kimi-K2-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "moonshotai/Kimi-K2-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/moonshotai/Kimi-K2-Instruct

SGLang

How to use moonshotai/Kimi-K2-Instruct 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 "moonshotai/Kimi-K2-Instruct" \
    --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": "moonshotai/Kimi-K2-Instruct",
		"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 "moonshotai/Kimi-K2-Instruct" \
        --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": "moonshotai/Kimi-K2-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use moonshotai/Kimi-K2-Instruct with Docker Model Runner:
```
docker model run hf.co/moonshotai/Kimi-K2-Instruct
```

Correct system prompt?

#28

by danielhanchen - opened Jul 18, 2025

Discussion

danielhanchen

Jul 18, 2025

Hey Moonshot team! Great work as usual!
I'm from Unsloth were we made 1.8bit quants for Kimi K2 🙂 https://huggingface.co/unsloth/Kimi-K2-Instruct-GGUF

I do have a question - do you guys know what is the "correct" system prompt?
The README.md mentions "You are Kimi, an AI assistant created by Moonshot AI."
But the tokenizer.json chat template uses "You are a helpful assistant"

According to https://platform.moonshot.ai/docs/guide/agent-support.en-US#direct-api-usage-for-kimi-k2-0711-preview-model it's "You are Kimi, an AI assistant provided by Moonshot AI. You are proficient in Chinese and English conversations. You provide users with safe, helpful, and accurate answers. You will reject any questions involving terrorism, racism, or explicit content. Moonshot AI is a proper noun and should not be translated."

Which is the correct recommended one? Thank you!

lychees

Jul 18, 2025

In most case, any system prompt should be fine. Or you are prefer to use the one in README.md to align the open source config.
Regards.

danielhanchen

Jul 18, 2025

•

edited Jul 18, 2025

@lychees It's cause sometimes the model might be trained with a designated system prompt, so it's always good to use whatever it was trained with.

Yulun

Jul 18, 2025

Normally, "You are Kimi, an AI assistant created by Moonshot AI." is good enough for most cases and reproducing benchmarks. If you are using our APIs, we recommend the longer one on our platform guide for safety reasons.

But really, K2 should follow any prompt. :D

danielhanchen

Jul 18, 2025

@Yulun Oh ok so essentially is "You are Kimi, an AI assistant created by Moonshot AI." better than the chat template default of "You are a helpful assistant"?

bigmoyan

Moonshot AI org Jul 18, 2025

@danielhanchen Hi , we have updated the default system prompt to "You are Kimi, an AI assistant created by Moonshot AI." in chat template. Thanks!

bigmoyan changed discussion status to closed Jul 18, 2025

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