Instructions to use XiaomiMiMo/MiMo-V2-Flash-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use XiaomiMiMo/MiMo-V2-Flash-Base with Transformers:

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

pipe = pipeline("text-generation", model="XiaomiMiMo/MiMo-V2-Flash-Base", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("XiaomiMiMo/MiMo-V2-Flash-Base", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("XiaomiMiMo/MiMo-V2-Flash-Base", trust_remote_code=True, device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use XiaomiMiMo/MiMo-V2-Flash-Base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "XiaomiMiMo/MiMo-V2-Flash-Base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "XiaomiMiMo/MiMo-V2-Flash-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/XiaomiMiMo/MiMo-V2-Flash-Base

SGLang

How to use XiaomiMiMo/MiMo-V2-Flash-Base 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 "XiaomiMiMo/MiMo-V2-Flash-Base" \
    --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": "XiaomiMiMo/MiMo-V2-Flash-Base",
		"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 "XiaomiMiMo/MiMo-V2-Flash-Base" \
        --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": "XiaomiMiMo/MiMo-V2-Flash-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use XiaomiMiMo/MiMo-V2-Flash-Base with Docker Model Runner:
```
docker model run hf.co/XiaomiMiMo/MiMo-V2-Flash-Base
```

nielsr HF Staff commited on Jan 8

Commit

ade85f0

verified ·

1 Parent(s): d852d0f

Add pipeline_tag and link to technical report

Browse files

Hi! I'm Niels, part of the community science team at Hugging Face.

This PR improves the model card for MiMo-V2-Flash by:
1. Adding the `pipeline_tag: text-generation` to the metadata for better discoverability.
2. Updating the technical report link to point to its official page on the Hugging Face Hub.
3. Explicitly linking the GitHub repository in the header for easier access to the code.

The rest of the high-quality documentation and structure provided by the Xiaomi team has been preserved.

Files changed (1) hide show

README.md +9 -4

README.md CHANGED Viewed

@@ -1,8 +1,9 @@
 ---
-license: mit
 base_model:
 - XiaomiMiMo/MiMo-V2-Flash-Base
 library_name: transformers
 ---
 <br/><br/>
@@ -20,10 +21,12 @@ library_name: transformers
   |
   <a href="https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash" target="_blank">🤗 HuggingFace</a>
   &nbsp;|
-  <a href="https://github.com/XiaomiMiMo/MiMo-V2-Flash/blob/main/paper.pdf" target="_blank">📔 Technical Report </a>
   &nbsp;|
   <a href="https://mimo.xiaomi.com/blog/mimo-v2-flash" target="_blank">📰 Blog </a>
   &nbsp;|
   <br/><br/>
   <strong>Play around!</strong> &nbsp;
   <a href="https://aistudio.xiaomimimo.com" target="_blank">🗨️ Xiaomi MiMo Studio </a>
@@ -36,6 +39,8 @@ library_name: transformers
 **MiMo-V2-Flash** is a Mixture-of-Experts (MoE) language model with **309B total parameters** and **15B active parameters**. Designed for high-speed reasoning and agentic workflows, it utilizes a novel hybrid attention architecture and Multi-Token Prediction (MTP) to achieve state-of-the-art performance while significantly reducing inference costs.
 <p align="center">
   <img width="80%" src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/MiMo-v2-flash-performance.jpg?raw=true">
 </p>
@@ -304,7 +309,7 @@ If you find our work helpful, please cite our technical report:
   title={MiMo-V2-Flash Technical Report},
   author={LLM-Core Xiaomi},
   year={2025},
-  url={https://github.com/XiaomiMiMo/MiMo-V2-Flash/paper.pdf}
 }
 ```
@@ -317,4 +322,4 @@ Please contact us at [mimo@xiaomi.com](mailto:mimo@xiaomi.com), join our WeChat
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat2.jpg?raw=true" width="20%" />
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat3.jpg?raw=true" width="20%" />
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat4.jpg?raw=true" width="20%" />
-</p>

 ---
 base_model:
 - XiaomiMiMo/MiMo-V2-Flash-Base
 library_name: transformers
+license: mit
+pipeline_tag: text-generation
 ---
 <br/><br/>
   |
   <a href="https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash" target="_blank">🤗 HuggingFace</a>
   &nbsp;|
+  <a href="https://huggingface.co/papers/2601.02780" target="_blank">📔 Technical Report </a>
   &nbsp;|
   <a href="https://mimo.xiaomi.com/blog/mimo-v2-flash" target="_blank">📰 Blog </a>
   &nbsp;|
+  <a href="https://github.com/XiaomiMiMo/MiMo-V2-Flash" target="_blank">💻 GitHub </a>
+  &nbsp;|
   <br/><br/>
   <strong>Play around!</strong> &nbsp;
   <a href="https://aistudio.xiaomimimo.com" target="_blank">🗨️ Xiaomi MiMo Studio </a>
 **MiMo-V2-Flash** is a Mixture-of-Experts (MoE) language model with **309B total parameters** and **15B active parameters**. Designed for high-speed reasoning and agentic workflows, it utilizes a novel hybrid attention architecture and Multi-Token Prediction (MTP) to achieve state-of-the-art performance while significantly reducing inference costs.
+The model was presented in the [MiMo-V2-Flash Technical Report](https://huggingface.co/papers/2601.02780).
 <p align="center">
   <img width="80%" src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/MiMo-v2-flash-performance.jpg?raw=true">
 </p>
   title={MiMo-V2-Flash Technical Report},
   author={LLM-Core Xiaomi},
   year={2025},
+  url={https://huggingface.co/papers/2601.02780}
 }
 ```
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat2.jpg?raw=true" width="20%" />
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat3.jpg?raw=true" width="20%" />
   <img src="https://github.com/XiaomiMiMo/MiMo-V2-Flash/raw/main/figures/wechat_group/wechat4.jpg?raw=true" width="20%" />
+</p>