arXiv:2608. 05381v1 Announce Type: new Abstract: Current Multimodal Large Language Models (MLLMs) can process diverse sensory inputs, yet their reasoning remains heavily biased toward a dominant modality, resulting in brittle cross-modal reasoning.
arXiv:2605.13737v2 Announce Type: replace Abstract: When an omnimodal large language model accepts a question whose textual premise contradicts what it actually sees or hears, does the failure lie in...
arXiv:2606. 00959v1 Announce Type: new Abstract: Understanding modality interaction in multimodal large language models (MLLMs) is central to reliable deployment.
arXiv:2609.06011v1 Announce Type: cross Abstract: Omni-modal large language models (OLLMs) jointly process vision, audio, and text, yet their modality bias under cross-modal conflict remains underexp...
arXiv:2609.18323v1 Announce Type: new Abstract: Recent Omni-Modal Generative Models (Omni-Models) have advanced content generation toward unified modeling of text, images, video, and audio. MiniMax-H...
OmniHallu is a unified framework for detecting hallucinations in multimodal large language models across both comprehension and generation tasks involving image, video, and audio modalities. It introduces OmniHallu-Bench, a 10,000-sample benchmark with claim-level human annotations for six cross-modal tasks (I2T, V2T, A2T, T2I, T2V, T2A). The system uses a multi‑agent architecture that decomposes outputs into atomic claims, verifies them with modality‑specific experts, and aggregates evidence through structured reasoning, while a preference‑optimized verifier reduces expert calls by 66% with minimal performance loss.
The paper investigates how multimodal large language models (MLLMs) handle conflicting evidence presented in text, image, or both forms. Across 13 MLLMs and two datasets, the authors find that models are not robust to knowledge conflict: they tend to accept contradictory image evidence more readily than contradictory text, and when both modalities conflict the preference is arbitrary, depending on input order, model, and dataset. The instability degrades multimodal retrieval-augmented generation and can be exploited by adversarial attacks, while simple mitigation techniques such as prompting, steering, and direct preference optimization largely fail, with supervised fine‑tuning offering only moderate improvement.
arXiv:2606. 05843v1 Announce Type: cross Abstract: While Multimodal Large Language Models (MLLMs) demonstrate remarkable proficiency on complex vision-language tasks, the mechanisms by which they extract query-relevant visual features from complex, noisy contexts remain opaque.
arXiv:2609.36798v1 Announce Type: cross Abstract: Omni-modal large language models (LLMs) are expected to answer a question using the modality it explicitly refers to. However, existing training para...
The Unwritten Benchmark introduces a novel challenge for multimodal machine learning, focusing on abstract perceptual reasoning through acousto‑kinematic word inference. Models must decode words written only by the audio of pen scratches and the video of hand movements, across three writing styles, without any visible ink trace. Evaluation shows a stark performance gap: humans achieve over 80% ordered letter accuracy, while leading models like GPT‑4o and Gemini 2.5‑Pro fail to exceed 10%, and combining modalities often degrades performance.
arXiv:2606. 09853v1 Announce Type: new Abstract: A central objective in multimodal learning is to capture synergy: task-relevant information that arises only from the joint use of multiple modalities, and is not available from any single modality alone.
arXiv:2608. 00076v2 Announce Type: replace-cross Abstract: Multimodal large language models (MLLMs) increasingly support high-stakes decision making by combining complementary information from images and text.
The paper investigates whether the layer that yields the highest probing accuracy in omni‑modal large language models is also the most effective for steering interventions. Across three independently developed models, the authors find that the best probing layers differ widely, whereas the most steerable layers consistently lie in a narrow mid‑to‑late range of the network. Using emotion as a testbed, they demonstrate a significant causal gap between probing and steering, and propose a two‑factor account linking readability and downstream plasticity to steering effectiveness.