Instructions to use meituan-longcat/LongCat-Flash-Lite with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use meituan-longcat/LongCat-Flash-Lite with Transformers:

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

pipe = pipeline("text-generation", model="meituan-longcat/LongCat-Flash-Lite", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("meituan-longcat/LongCat-Flash-Lite", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use meituan-longcat/LongCat-Flash-Lite with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/meituan-longcat/LongCat-Flash-Lite

SGLang

How to use meituan-longcat/LongCat-Flash-Lite 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 "meituan-longcat/LongCat-Flash-Lite" \
    --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": "meituan-longcat/LongCat-Flash-Lite",
		"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 "meituan-longcat/LongCat-Flash-Lite" \
        --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": "meituan-longcat/LongCat-Flash-Lite",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use meituan-longcat/LongCat-Flash-Lite with Docker Model Runner:
```
docker model run hf.co/meituan-longcat/LongCat-Flash-Lite
```

使用sglang在两个H200上推理速度非常慢

by taozi555 - opened Feb 2

Discussion

taozi555

Feb 2

python3 -m sglang.launch_server
--model meituan-longcat/LongCat-Flash-Lite
--port 6006
--host 0.0.0.0
--mem-fraction-static 0.9
--max-running-requests 64
--trust-remote-code
--skip-server-warmup
--attention-backend flashinfer
--ep 2
--tp 2
--disable-cuda-graph

使用此命令，大约12tk/s

malithh

Feb 3

我们是 DeployPad。
我们注意到使用 sglang 在 2× H200 上运行 LongCat 只有约 12 tok/s，这一性能明显偏低。

👉DeployPad 推理栈将于本周末正式上线
👉 上线即支持 LongCat：单张 H200 可达约 60–80 tok/s，同时支持 RTX Pro 6000
👉 将向 LongCat 社区开放支持

我们的目标是充分发挥 H200 的性能，在无需复杂调参的情况下实现更高吞吐。

周末见。

sjqgogogogo

Feb 28

Hi, we now support cuda graph on sglang via PR https://github.com/sgl-project/sglang/pull/17838

With cuda graph on, you can run LongCat-Flash-Lite-FP8 on H800*8 at about 250TPS/user

python3 -m sglang.launch_server \
    --model meituan-longcat/LongCat-Flash-Lite-FP8 \
    --port 8080 \
    --host 0.0.0.0 \
    --mem-fraction-static 0.9 \
    --max-running-requests 64 \
    --trust-remote-code \
    --skip-server-warmup \
    --attention-backend flashinfer \
    --ep 8 \
    --tp 8 \
    --cuda-graph-bs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 20 24 28 32 36 40 44 48 52 56 60 64

Additionally, for the best possible performance, we highly recommend trying out our inference engine, SGLang-FluentLLM. You can find more details in https://github.com/meituan-longcat/SGLang-FluentLLM/tree/main

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