Modelpublic

stepfun-ai/step3

sourceHugging Faceapache-2.0updated 8mo agoView on Hugging Face

166likes37kdownloads

Model Card

<hr>

<div align="center"> <b>📰  <a href="https://stepfun.ai/research/step3">Step3 Model Blog</a></b>     |     <b>📄  <a href="https://arxiv.org/abs/2507.19427">Step3 System Blog</a></b> </div>

Introduction

Step3 is our cutting-edge multimodal reasoning model—built on a Mixture-of-Experts architecture with 321B total parameters and 38B active. It is designed end-to-end to minimize decoding costs while delivering top-tier performance in vision–language reasoning. Through the co-design of Multi-Matrix Factorization Attention (MFA) and Attention-FFN Disaggregation (AFD), Step3 maintains exceptional efficiency across both flagship and low-end accelerators.

Step3 model card:

Config	Value
Number of Layers (Dense layer included)	61
Number of Dense Layers	5
Hidden Dimension	7168
Attention Mechanism	MFA
Low-rank Query Dimension	2048
Number of Query Heads	64
Head Dimension	256
Number of Experts	48
Selected Experts per Token	3
Number of Shared Experts	1
Max Context Length	65536
Tokenizer	Deepseek V3
Total Parameters (LLM)	316B
Activated Params per Token	38B
Total Parameters (VLM)	321B

Evaluation Results

[image]

Deployment

[!Note] Step3's API is accessible at https://platform.stepfun.com/, where we offer OpenAI-compatible API for you.

Inference with Hugging Face Transformers

We introduce how to use our model at inference stage using transformers library. It is recommended to use python=3.10, torch>=2.1.0, and transformers=4.54.0 as the development environment.We currently only support bf16 inference, and multi-patch for image preprocessing is supported by default. This behavior is aligned with vllm and sglang.

python

from transformers import AutoProcessor, AutoModelForCausalLM

key_mapping = {
    "^vision_model": "model.vision_model",
    r"^model(?!\.(language_model|vision_model))": "model.language_model",
    "vit_downsampler": "model.vit_downsampler",
    "vit_downsampler2": "model.vit_downsampler2",
    "vit_large_projector": "model.vit_large_projector",
}

model_path = "stepfun-ai/step3"

processor = AutoProcessor.from_pretrained(model_path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_path, 
                device_map="auto", torch_dtype="auto",trust_remote_code=True, 
                key_mapping=key_mapping)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "What's in this picture?"}
        ]
    },
]

inputs = processor.apply_chat_template(
    messages, add_generation_prompt=True, tokenize=True,
    return_dict=True, return_tensors="pt"
).to(model.device)

generate_ids = model.generate(**inputs, max_new_tokens=32768, do_sample=False)
decoded = processor.decode(generate_ids[0, inputs["input_ids"].shape[-1] :], skip_special_tokens=True)

print(decoded)

Inference with vLLM and SGLang

Our model checkpoints are stored in bf16 and block-fp8 format, you can find it on Huggingface.

Currently, it is recommended to run Step3 on the following inference engines:

—vLLM
—SGLang

Deployment and Request examples for vLLM and SGLang can be found in the Model Deployment Guide.

Contact Us

If you have any questions, please reach out at contact@stepfun.com .

License

Both the code repository and the model weights are released under the Apache License (Version 2.0).

Citation

@misc{step3system,
      title={Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decoding}, 
      author={StepFun Team},
      year={2025},
      eprint={2507.19427},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2507.19427}, 
}

@misc{step3blog,
      title={Step3: Cost-Effective Multimodal Intelligence}, 
      author={StepFun Team},
      url={https://stepfun.ai/research/step3}, 
}