Instructions to use moonshotai/Kimi-K2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use moonshotai/Kimi-K2.5 with Transformers:

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

pipe = pipeline("image-text-to-text", model="moonshotai/Kimi-K2.5", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("moonshotai/Kimi-K2.5", trust_remote_code=True, dtype="auto")

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use moonshotai/Kimi-K2.5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "moonshotai/Kimi-K2.5"
# 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-K2.5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/moonshotai/Kimi-K2.5

SGLang

How to use moonshotai/Kimi-K2.5 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-K2.5" \
    --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-K2.5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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-K2.5" \
        --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-K2.5",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use moonshotai/Kimi-K2.5 with Docker Model Runner:
```
docker model run hf.co/moonshotai/Kimi-K2.5
```

Should `rope_scaling.beta_fast` be `1.0`?

by jukofyork - opened Jan 27

Discussion

jukofyork

Jan 27

The previous K2 models have used 1.0 for this value and it appears to have been a copy and paste bug from the deepseek3 architecture in the past:

https://github.com/radixark/miles/issues/335

Does v2.5 use 32.0 or is this a config bug that needs fixing?

bigeagle

Moonshot AI org Jan 27

K2.5 uses rope_scaling.beta_fast=32.0 :)

jukofyork

Jan 27

K2.5 uses rope_scaling.beta_fast=32.0 :)

Oh thanks! Will close now.

jukofyork changed discussion status to closed Jan 27

jukofyork

Jan 30

•

edited Jan 30

I've been investigating some more and it seems the original Kimi models may have used the wrong value for rope_scaling.beta_fast=1.0 as the paper says:

but when I investigated why this doesn't seem to effect llama.cpp:

https://github.com/ggml-org/llama.cpp/blob/83bcdf7217dc06ac67ff5f7322bdd89f46664c04/src/llama-cparams.h#L22

Not sure if this comment is out of date, but it looks like it is actually using a fixed value of 32.0.

It seems the same is true for ik_llama.cpp too:

https://github.com/ikawrakow/ik_llama.cpp/blob/811f8c339373b0d5f9c30d7021fe3d66629b2750/src/llama-cparams.h#L22

jukofyork changed discussion status to open Jan 30

jukofyork

Jan 30

@bigeagle can you confirm if the older Kimi models should really have used rope_scaling.beta_fast=1.0 or if it was a config bug and should have been rope_scaling.beta_fast=32.0?

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