MATEO: A Multimodal Benchmark for Temporal Reasoning and Planning in LVLMs

TL;DR

MATEO is a new benchmark designed to evaluate and enhance the ability of large vision-language models to understand and reason about complex, temporally ordered multimodal instructions for real-world planning tasks.

Contribution

The paper introduces MATEO, a comprehensive multimodal benchmark with a high-quality recipe corpus and TEO annotations, enabling systematic assessment of LVLMs' temporal reasoning capabilities.

Findings

LVLMs show varying performance on temporal reasoning tasks

Fine-tuning strategies impact model understanding of temporal order

Multimodal input structure influences reasoning accuracy

Abstract

AI agents need to plan to achieve complex goals that involve orchestrating perception, sub-goal decomposition, and execution. These plans consist of ordered steps structured according to a Temporal Execution Order (TEO, a directed acyclic graph that ensures each step executes only after its preconditions are satisfied. Existing research on foundational models' understanding of temporal execution is limited to automatically derived annotations, approximations of the TEO as a linear chain, or text-only inputs. To address this gap, we introduce MATEO (MultimodAl Temporal Execution Order), a benchmark designed to assess and improve the temporal reasoning abilities of Large Vision Language Models (LVLMs) required for real-world planning. We acquire a high-quality professional multimodal recipe corpus, authored through a standardized editorial process that decomposes instructions into discrete steps, each paired with corresponding images. We collect TEO annotations as graphs by designing and using a scalable crowdsourcing pipeline. Using MATEO, we evaluate six state-of-the-art LVLMs across model scales, varying language context, multimodal input structure, and fine-tuning strategies.