STEP3-VL-10B 技术报告
计算机视觉与模式识别
2026-01-16 v2
摘要
我们提出 STEP3-VL-10B，一个轻量级开源基础模型，旨在重新定义紧凑效率与前沿多模态智能之间的权衡。STEP3-VL-10B 通过两个战略转变实现：首先，在 1.2T 多模态 token 上采用统一的、完全解冻的预训练策略，将语言对齐的感知编码器与 Qwen3-8B 解码器集成，以建立内在的视觉-语言协同；其次，一个包含超过 1000 次强化学习迭代的规模化后训练流程。至关重要的是，我们实现了并行协调推理 (PaCoRe) 以扩展测试时计算，将资源分配给可扩展的感知推理，探索并综合多样化的视觉假设。因此，尽管 STEP3-VL-10B 仅有 100 亿的紧凑规模，但其性能媲美或超越了比其大 10 到 20 倍的模型（例如 GLM-4.6V-106B、Qwen3-VL-235B）以及顶级专有旗舰模型如 Gemini 2.5 Pro 和 Seed-1.5-VL。它提供了同类最佳的性能，在 MMBench 上达到 92.2%，在 MMMU 上达到 80.11%，同时在复杂推理方面表现出色，在 AIME2025 上达到 94.43%，在 MathVision 上达到 75.95%。我们发布完整的模型套件，为社区提供一个强大、高效且可复现的基线。
引用
@article{arxiv.2601.09668,
title = {STEP3-VL-10B Technical Report},
author = {Ailin Huang and Chengyuan Yao and Chunrui Han and Fanqi Wan and Hangyu Guo and Haoran Lv and Hongyu Zhou and Jia Wang and Jian Zhou and Jianjian Sun and Jingcheng Hu and Kangheng Lin and Liang Zhao and Mitt Huang and Song Yuan and Wenwen Qu and Xiangfeng Wang and Yanlin Lai and Yingxiu Zhao and Yinmin Zhang and Yukang Shi and Yuyang Chen and Zejia Weng and Ziyang Meng and Ang Li and Aobo Kong and Bo Dong and Changyi Wan and David Wang and Di Qi and Dingming Li and En Yu and Guopeng Li and Haiquan Yin and Han Zhou and Hanshan Zhang and Haolong Yan and Hebin Zhou and Hongbo Peng and Jiaran Zhang and Jiashu Lv and Jiayi Fu and Jie Cheng and Jie Zhou and Jisheng Yin and Jingjing Xie and Jingwei Wu and Jun Zhang and Junfeng Liu and Kaijun Tan and Kaiwen Yan and Liangyu Chen and Lina Chen and Mingliang Li and Qian Zhao and Quan Sun and Shaoliang Pang and Shengjie Fan and Shijie Shang and Siyuan Zhang and Tianhao You and Wei Ji and Wuxun Xie and Xiaobo Yang and Xiaojie Hou and Xiaoran Jiao and Xiaoxiao Ren and Xiangwen Kong and Xin Huang and Xin Wu and Xing Chen and Xinran Wang and Xuelin Zhang and Yana Wei and Yang Li and Yanming Xu and Yeqing Shen and Yuang Peng and Yue Peng and Yu Zhou and Yusheng Li and Yuxiang Yang and Yuyang Zhang and Zhe Xie and Zhewei Huang and Zhenyi Lu and Zhimin Fan and Zihui Cheng and Daxin Jiang and Qi Han and Xiangyu Zhang and Yibo Zhu and Zheng Ge},
journal= {arXiv preprint arXiv:2601.09668},
year = {2026}
}
备注
50 pages