Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models

TL;DR

Reinforcement fine-tuning significantly enhances the reasoning abilities of multimodal large language models, advancing towards artificial general intelligence by integrating diverse modalities, tasks, and improved training methods.

Contribution

This paper argues that reinforcement fine-tuning is key to improving reasoning in multimodal large language models and summarizes recent progress and future directions in this area.

Findings

RFT improves reasoning across multiple modalities and tasks.

Enhanced training algorithms and benchmarks support RFT development.

The community has developed frameworks facilitating RFT for MLLMs.

Abstract

Standing in 2025, at a critical juncture in the pursuit of Artificial General Intelligence (AGI), reinforcement fine-tuning (RFT) has demonstrated significant potential in enhancing the reasoning capability of large language models (LLMs) and has led to the development of cutting-edge AI models such as OpenAI-o1 and DeepSeek-R1. Moreover, the efficient application of RFT to enhance the reasoning capability of multimodal large language models (MLLMs) has attracted widespread attention from the community. In this position paper, we argue that reinforcement fine-tuning powers the reasoning capability of multimodal large language models. To begin with, we provide a detailed introduction to the fundamental background knowledge that researchers interested in this field should be familiar with. Furthermore, we meticulously summarize the improvements of RFT in powering reasoning capability of MLLMs into five key points: diverse modalities, diverse tasks and domains, better training algorithms, abundant benchmarks and thriving engineering frameworks. Finally, we propose five promising directions for future research that the community might consider. We hope that this position paper will provide valuable insights to the community at this pivotal stage in the advancement toward AGI. Summary of works done on RFT for MLLMs is available at https://github.com/Sun-Haoyuan23/Awesome-RL-based-Reasoning-MLLMs.