Omni-R1: Towards the Unified Generative Paradigm for Multimodal Reasoning

TL;DR

This paper introduces Omni-R1, a unified generative framework for multimodal reasoning that generates intermediate images to handle diverse tasks, with a zero-shot variant that reduces annotation needs.

Contribution

The paper proposes a novel unified generative paradigm for multimodal reasoning and instantiates it with Omni-R1, including a zero-shot version that bypasses multimodal annotations.

Findings

Omni-R1 achieves unified reasoning across various multimodal tasks.

Omni-R1-Zero matches or surpasses Omni-R1 without multimodal annotations.

Empirical results demonstrate the effectiveness of the proposed approach.

Abstract

Multimodal Large Language Models (MLLMs) are making significant progress in multimodal reasoning. Early approaches focus on pure text-based reasoning. More recent studies have incorporated multimodal information into the reasoning steps; however, they often follow a single task-specific reasoning pattern, which limits their generalizability across various multimodal tasks. In fact, there are numerous multimodal tasks requiring diverse reasoning skills, such as zooming in on a specific region or marking an object within an image. To address this, we propose unified generative multimodal reasoning, which unifies diverse multimodal reasoning skills by generating intermediate images during the reasoning process. We instantiate this paradigm with Omni-R1, a two-stage SFT+RL framework featuring perception alignment loss and perception reward, thereby enabling functional image generation. Additionally, we introduce Omni-R1-Zero, which eliminates the need for multimodal annotations by bootstrapping step-wise visualizations from text-only reasoning data. Empirical results show that Omni-R1 achieves unified generative reasoning across a wide range of multimodal tasks, and Omni-R1-Zero can match or even surpass Omni-R1 on average, suggesting a promising direction for generative multimodal reasoning.