Newest papers

OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory

Yu Zhou, Cheng-Fu Yang, Rui Sun, Di Wu, Sibo Peng, Mingyang Zhang, Zhen Yang, Ruiyi Zhang, Nanyun Peng, Zhe Gan, Kai‐Wei Chang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26860 · cited by 0 works (OpenAlex) · open-access PDF

Russian

Мультимодальные большие языковые модели (MLLMs) сделали быстрый прогресс в визуальном рассуждении, но многие задачи требуют доказательств, которые трудно получить только с помощью внутреннего рассуждения: наше контролируемое исследование по 22 этапам их рассуждения показывает, что простое увеличение бюджета вывода базовой модели приносит только скромные выгоды. Такие задачи требуют наземного восприятия, внешних знаний и численных измерений, полученных с помощью гетерогенных инструментов, в то время как выходы которых должны быть произведены максимально интерпретируемыми для модели и сохранены на этапах открытого рассуждения. Мы вводим OpenVLHarness, мультимодальную рассудочную гамму, организованную вокруг слоя возможностей и

Machine translation (Meta M2M100); the English original follows.

Abstract

Multimodal large language models (MLLMs) have made rapid progress on visual reasoning, yet many tasks require evidence that is difficult to obtain through internal reasoning alone: our controlled study on 22 benchmarks shows that simply increasing the inference budget of the base model yields only modest gains. Such tasks demand grounded perception, external knowledge, and numerical measurements obtained through heterogeneous tools, whose outputs must be made interpretable to the model and retained across reasoning steps. We introduce OpenVLHarness, a multimodal reasoning harness organized around a capability layer and persistent multimodal memory. The capability layer combines task-level tool interfaces with evidence rendering, providing consistent access to heterogeneous backends and converting their outputs into model-readable observations. Persistent multimodal memory retains the corresponding images, structured data, and web references as addressable artifacts throughout each query. A coding interface executes generated programs over stored data, enabling computations that combine results from multiple tools. Together, these mechanisms support an iterative process of acquiring, inspecting, and composing multimodal evidence. Evaluated on 22 benchmarks with eight open and proprietary backbones, OpenVLHarness improves average scores by 9.7-12.6 points over matched no-tool baselines, including +11.2 with Qwen3-VL-8B, +9.7 with Kimi K3, and +9.9 with GPT-6 Astra. With GPT-6 Luna at medium reasoning effort, OpenVLHarness outperforms the base model, Codex, and Visual Sketchpad at maximum effort by 7.7-8.5 points while incurring at most 37% of their model-token cost.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Coding (social sciences), Inference, Visual reasoning, Multimodality, Process (computing)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange