Vision-language models (VLMs) have advanced rapidly in multimodal reasoning, yet recent work shows that their failures often reflect an interaction between visual grounding and downstream reasoning. What remains less clear is how the visual presentation of a task shapes model performance and failure modes when the underlying reasoning problem is unchanged. We study this question in SPaRC, a benchmark for grid-based visual spatial planning, by introducing lightweight input-side scaffolds that preserve the visual modality while making spatial structure more accessible. The scaffolds are task-specific diagnostic interventions built on SPaRC’s known board structure, not a general-purpose prompting strategy. Across multiple VLMs, these scaffolds improve task accuracy over the original visual setting by up to 34.0 percentage points and further complement GRPO-based training, yielding up to 4.6 additional accuracy points compared with near-zero gains on the original visual input. Analyses on both end-to-end task solving and object detection show that these gains are closely tied to reductions in grounding-related errors, while rule reasoning remains comparatively challenging. With the planning problem held fixed, visual presentation therefore changes both accuracy and the distribution of failure modes, which in turn determines whether a VLM benchmark measures grounded perception, downstream reasoning, or a mixture of both.

We present DIN 19461:2026-06 (E), a German draft national standard that defines categories, terminology, and process requirements for Derived Text Formats (DTFs) created from text documents in natural language. The standard specifies enrichment and information reduction operations, requirements for combining multiple DTFs, and documentation obligations for publication, archiving, and reuse. Its aim is to enable legally compliant sharing and analysis of texts–especially where copyright or data protection prevents distributing originals–while maintaining scientific utility and reproducibility through explicit process and parameter recording. We outline the scope, the key concepts, the four core reduction operations (retain, delete, replace, randomise), together with examples across token-, structure-, and vector-based DTFs, and implications for infrastructures (e.g., ISO 24622-based metadata). Finally, we discuss limitations, open questions (e.g., reconstruction risks with modern ML models), and next steps for adoption and maintenance.

While Vision–language models (VLMs) interpret text-rich images effectively, they struggle with reasoning across long, multi-page documents. We present Active 𝐋ong 𝐃ocum𝐄nt 𝐍avigation (ALDEN), a multi-turn reinforcement learning framework that fine-tunes VLMs as interactive agents capable of actively navigating long, visually rich documents rather than passive readers. ALDEN features a novel fetch action that allows direct page indexing, complementing the classic search action and better exploiting document structure. To ensure training efficiency and stability, we introduce a rule-based cross-level reward for dense supervision and a visual-semantic anchoring mechanism utilizing dual-path KL-divergence constraints. We train ALDEN on a curated corpus built from open-source datasets where trivial samples are filtered, and queries are rewritten to incentivize multi-turn navigation and fetch usage. Empirically, ALDEN achieves state-of-the-art results on five long-document benchmarks, offering a more accurate and efficient path for long-document understanding.