ROVER: Benchmarking Reciprocal Cross-Modal Reasoning for Omnimodal Generation

TL;DR

ROVER is a benchmark designed to evaluate reciprocal cross-modal reasoning in unified multimodal models, addressing the gap in current assessments by testing how models use one modality to guide or verify outputs in the other.

Contribution

It introduces a human-annotated benchmark with 1312 tasks for reciprocal cross-modal reasoning, spanning visual and verbal modalities, to evaluate and improve unified multimodal models.

Findings

Interleaved models outperform non-interleaved ones in cross-modal reasoning.

Cross-modal reasoning quality impacts visual generation performance.

Models excel at literal perceptual concepts but struggle with visual abstractions for symbolic reasoning.

Abstract

Unified multimodal models (UMMs) have emerged as a powerful paradigm for seamlessly unifying text and image understanding and generation. However, prevailing evaluations treat these abilities in isolation, such that tasks with multimodal inputs and outputs are scored primarily through unimodal reasoning, i.e., textual benchmarks emphasize language-based reasoning, while visual benchmarks emphasize reasoning outcomes manifested in the pixels. We introduce ROVER to address this pressing need to test reciprocal cross-modal reasoning, the use of one modality to guide, verify, or refine outputs in the other, an ability central to the vision of unified multimodal intelligence. ROVER is a human-annotated benchmark that explicitly targets reciprocal cross-modal reasoning, which contains 1312 tasks grounded in 1876 images, spanning two complementary settings. Verbally-augmented reasoning for visual generation evaluates whether models can use verbal prompts and reasoning chains to guide faithful image synthesis. Visually-augmented reasoning for verbal generation evaluates whether models can generate intermediate visualizations that strengthen their own reasoning processes for question answering. Experiments on 17 unified models reveal two key findings: (i) Cross-modal reasoning determines visual generation quality, with interleaved models significantly outperforming non-interleaved ones; notably, combining strong unimodal models fails to achieve comparable reasoning. (ii) Models show dissociation between physical and symbolic reasoning: they succeed at interpreting perceptual concepts literally but fail to construct visual abstractions for symbolic tasks, where faulty reasoning harms performance. These results highlight reciprocal cross-modal reasoning as a critical frontier for enabling true omnimodal generation.