Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
Abstract
We introduce InternVL 2.5, an advanced multimodal large language model (MLLM) series that builds upon InternVL 2.0, maintaining its core model architecture while introducing significant enhancements in training and testing strategies as well as data quality. In this work, we delve into the relationship between model scaling and performance, systematically exploring the performance trends in vision encoders, language models, dataset sizes, and test-time configurations. Through extensive evaluations on a wide range of benchmarks, including multi-discipline reasoning, document understanding, multi-image / video understanding, real-world comprehension, multimodal hallucination detection, visual grounding, multilingual capabilities, and pure language processing, InternVL 2.5 exhibits competitive performance, rivaling leading commercial models such as GPT-4o and Claude-3.5-Sonnet. Notably, our model is the first open-source MLLMs to surpass 70% on the MMMU benchmark, achieving a 3.7-point improvement through Chain-of-Thought (CoT) reasoning and showcasing strong potential for test-time scaling. We hope this model contributes to the open-source community by setting new standards for developing and applying multimodal AI systems. HuggingFace demo see https://huggingface.co/spaces/OpenGVLab/InternVL
Cite
@article{arxiv.2412.05271,
title = {Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling},
author = {Zhe Chen and Weiyun Wang and Yue Cao and Yangzhou Liu and Zhangwei Gao and Erfei Cui and Jinguo Zhu and Shenglong Ye and Hao Tian and Zhaoyang Liu and Lixin Gu and Xuehui Wang and Qingyun Li and Yiming Ren and Zixuan Chen and Jiapeng Luo and Jiahao Wang and Tan Jiang and Bo Wang and Conghui He and Botian Shi and Xingcheng Zhang and Han Lv and Yi Wang and Wenqi Shao and Pei Chu and Zhongying Tu and Tong He and Zhiyong Wu and Huipeng Deng and Jiaye Ge and Kai Chen and Kaipeng Zhang and Limin Wang and Min Dou and Lewei Lu and Xizhou Zhu and Tong Lu and Dahua Lin and Yu Qiao and Jifeng Dai and Wenhai Wang},
journal= {arXiv preprint arXiv:2412.05271},
year = {2025}
}
Comments
Technical Report