Compositional Reasoning in VLMs

Relational out-of-distribution errors are easy to describe and hard to evaluate cleanly. An image may be natural, a caption may be fluent, and all mentioned objects may be common, yet the caption may describe the wrong relation in the image. Many evaluations intended to test this behavior can be solved without checking the pairing at all, because negatives differ in caption style, template, length, object frequency, or image statistics. We propose counterfactual pairing cycles, a blocked evaluation primitive that reuses the same two images and the same two captions on both sides of the comparison, changing only which image is paired with which caption. This makes image-only, text-only, and additive component shortcuts cancel exactly in finite samples. The same contrast also gives a training loss for relational outlier exposure and a diagnostic for bilinear vision-language scores. Across synthetic controls, GQA, and COCO, cycles separate coarse image-caption matching from directional role-binding failures, while also identifying geometric size relations that remain strongly visible in global features. In a shortcut-contaminated COCO outlier-exposure stress test, a model selected by a near-perfect shortcut validation AUROC retains substantial relational signal but remains below cycle training on exact cycles; cycle training improves CycleAcc from 0.882 to 0.927. The main lesson is simple: to test whether a model understands a pairing, the benchmark should keep the parts fixed, change the pairing, and ask only about coupling. The official implementation is available online.

Vision-language models (VLMs) increasingly reason over visual evidence that is cropped, segmented, retrieved, or revealed over time. Yet most VQA benchmarks present the complete image and question at once. We ask what models lose when the same information is fragmented. We introduce Layered-VQA, with 93 scenes and 300 questions. Each image is decomposed into ordered RGBA layers that exactly recompose the original scene, and each question is annotated with supporting, minimal-sufficient, and distractor layers. We evaluate eleven open-weight VLMs from 3B to 32B parameters and two proprietary models with a scale of 187,200 conversations, graded by 1.74M open-model cross-judgments. We find three consistent failures. Loss in Composition: fragmenting the question has a small effect, but fragmenting the scene substantially reduces accuracy; recomposing the same layers largely restores performance. Oracle Inversion: even oracle-selected sufficient evidence can perform worse than the complete scene. Loss in Grounding: as more evidence is required, grounding degrades much faster than answer accuracy. Together, these results show that having the right visual evidence is not enough. How that evidence is composed and presented determines whether models can use and ground it. The right evidence is not enough: VLMs need the scene it came from.

Vision-language models (VLMs) can answer simple visual questions, but often struggle when one question requires several visual judgments. We study this gap with controlled tasks for feature binding, numerosity, spatial relations, and amodal completion, together with a Composite task that combines them. Matched counterfactual image pairs isolate changes in the visual evidence needed to answer. Across four models, direct answers, hidden-state readouts, and state interventions show that the individual judgments can be made without explicit reasoning and that intervening on the corresponding states can affect the answer. During reasoning, the Composite answer becomes decodable from hidden states and usable from shortened traces, often before the model stops on its own. We train a small detector to predict this readiness and stop reasoning at that point. On MMStar and RealWorldQA, this reduces mean reasoning tokens by 79.1% and 74.5%, while average accuracy rises by 3.13 and 3.30 percentage points, respectively. These findings connect the internal development of answer readiness to a practical rule for allocating reasoning computation.

Vision-language models (VLMs) often struggle with compositional reasoning tasks, but the reasons for this underperformance remain unclear. A common hypothesis is that models struggle to integrate multiple components, leading to training interventions to improve compositional binding. However, this assumption has never been directly quantified. Existing benchmarks evaluate captions only in their composed form, making it impossible to separate the cost of joint reasoning from the cost of recognizing individual components under increasing load. We introduce COMPASS (COMPositional Analysis of SkillS), a controlled evaluation framework designed to isolate and measure the distinct factors underlying compositional failure. By comparing performance on composed captions with their decomposed counterparts , we directly quantify the cost of compositional integration across 87K image-caption pairs. Across multiple VLMs, this gap is real but partial, accounting for only part of the observed degradation. This motivates a finer-grained investigation into what additional factors govern model behavior. We analyze performance at the level of individual skills: object detection, attribute binding, and relation reasoning, using skill-targeted perturbations across 274K image-caption pairs. We find a consistent skill-specific pattern: each skill degrades primarily with the count of its own primitive type (self-load), while cross-load effects are predominantly positive, suggesting that primitives of different types provide useful grounding context. This pattern holds across standard contrastive encoders, explicitly trained compositional reasoning models, and non-contrastive architectures. These findings show that compositional degradation reflects multiple separable factors that cannot be reduced to joint reasoning alone.

Vision-language models (VLMs) demonstrate strong performance across compositional reasoning benchmarks, which require reasoning over semantic perturbations of objects, attributes, relations, and their interactions. However, our controlled analysis reveals that existing compositionality-aware VLMs exhibit element-specific biases, often underperforming vanilla CLIP on certain compositional elements. To address this, we propose Compositional Scene Graph-guided CLIP (CS-CLIP), which uses scene graphs to identify compositional elements and construct structured negatives via selective masking. We further retain negatives that are most contradictory to the original caption, forcing the model to rely on compositional structure rather than surface cues. CS-CLIP achieves state-of-the-art compositional reasoning with robust performance across compositional elements. It also preserves general vision-language capabilities such as cross-modal retrieval and downstream visual reasoning, while requiring fewer training samples than prior methods.

Dual-encoder models such as CLIP score an image-caption pair by a single inner product of two independently computed unit vectors, and fail at binding, often scoring near chance when asked to distinguish "a red car and a blue dog" from "a blue car and a red dog". We give a mathematical account of when this failure is necessary and when it is contingent. Working within the ideal-encoder framework proposed by Kang et al., we first show the relevant axioms are satisfiable, so every impossibility must enter through an added, checkable hypothesis. We then prove three such obstructions. Depth: for recursive role-binding codes the swap margin obeys an exact law $m(D) = 2b^{-D}$ in the nesting depth D, with a finite-dimension version holding up to one explicitly flagged concentration estimate; the resolvable depth grows only logarithmically in the dimension and is single-digit at CLIP scale, the nesting depth of ordinary language. Objective: architecture-free throttle theorems showing that the contrastive objective's entire reward for binding is bounded by the rate at which training contrasts a caption against its own swap, a rate that vanishes at web scale, and that exactly reversed binding costs only that rate times the mean binding margin; both are verified in simulation. Geometry: a tight smoothness-binding frontier: the closer the two swap-related captions must embed to a shared paraphrase anchor, the smaller the binding margin can be, with an exact constant. Measuring its text-only diagnostic across 18 deployed text encoders, every model sits at roughly 25-35% of its ceiling, and the induced per-item ceiling tracks SugarCrepe's subset difficulty at r = 0.99. Binding failure in deployed dual encoders is thus not a dimension or smoothness limit today, but an incentive and code-structure limit, with a proved depth ceiling that remains once those are fixed.

We build VectraYX-Vision-1B, a sub-2B Spanish/LATAM cybersecurity vision-language model coupling a frozen SigLIP-so400m encoder to a 1.04B-parameter decoder via a two-layer MLP projector, and report a diagnostic negative result: not that visual grounding failed, but why. After repairing five silent fine-tuning defects, grounding on a nine-field extraction gate with a shuffled-image control is 2/9, invariant across every configuration that leaves the encoder alone; 2x2 tiling, the one that changes it, loses a field and gains none. Resolution is not the operative variable: the field read almost perfectly has the highest entropy in the corpus. A linear probe on frozen SigLIP features gives per-glyph recoverability p~0.61, predicting 1.9% against an observed 0.00; tiling nearly doubles recoverability on two fields, yet the end-to-end model gets worse. Transplanting a natively-trained visual tower onto the same frozen decoder and recipe takes that address field from 0.00 to 0.81 exact, on a coarser token budget than the tiling condition that recovered nothing: pretraining regime, not resolution, sets how far the losses reach. A later, separately trained checkpoint adds one positive result: on B8 (34 fields, 16 templates, 2,040 items, dual shuffled-image/best-constant control), 9 fields pass, confirming genuine grounding within trained template-field combinations only. Sharpest new finding: inside a well-trained template, an untrained field returns a near-constant wrong answer independent of the image -- landmark-keyed lookup, not free-text reading. B6/B7 tool identification stays at 0.0 on every checkpoint including this one; we retract an earlier 0.08 tool-id score after finding three harness defects a stronger model would conceal. We release code, all three benchmarks, configs, and all training checkpoints, including the B8 corpus.

Dual-encoder vision-language models (VLMs) expose a similarity interface that enables zero-shot retrieval but fails compositional constraints: queries like "umbrella and no person" retrieve images containing both, even when concept detection is reliable. We trace this to an interface-level Bag-of-Concepts effect, where similarity scores approximate mean pooling of concept evidence regardless of operators. Although operator-dependent signals exist in text embeddings, they are too weak or misaligned to affect rankings. Fine-tuning does not reliably resolve this failure because the dominant bottleneck is how similarity aggregates evidence rather than what encoders represent. We propose factored inference, which separates evidence extraction from constraint execution, and introduce LCSE (Logic-Constrained Score Editing), a training-free method that executes constraints externally using concept scores from frozen encoders. We also introduce FACTOR-Bench, where LCSE achieves 85.5% accuracy versus 73.2% for the best fine-tuned baseline, 90.7% when applied to SigLIP 2, and improves NegBench COCO MCQ accuracy from 27.2% to 65.2% while preserving retrieval performance.

Vision-Language Models (VLMs) often achieve high performance on benchmarks while remaining "black boxes", yet they remain prone to hallucination or rely on superficial shortcuts. In this work, we propose a framework designed to enhance both performance and interpretability through De-compositional Evidence Grounding. Unlike monolithic inference approaches, our approach forces the model to decompose a global query into a sequence of atomic sub-questions, each requiring an explicit sub-answer and critically a localized evidence bounding box. By grounding intermediate logical steps (e.g. identifying a container, analyzing liquid properties, and assessing environmental context) in specific visual regions, we construct a structured reasoning path that mirrors human-like deduction. This allows the final answer to emerge as a logical consequence of verified visual facts rather than a statistical guess.

Conventional vision-language models are largely object-centric, focusing on detecting and describing individual entities. In safety-critical X-ray baggage screening, however, threat often emerges not from a single object but from the functional compatibility of spatially dispersed components, such as batteries, detonators, and explosive charges. We formalize this setting as compositional threat reasoning, where risk is modeled as a relational property of grounded regions rather than an independent detection outcome. We introduce Falcon, a multimodal framework that abstracts segmentation-aware region features into a structured safety state capturing component presence, pairwise functional compatibility, and scene-level risk. This structured representation is injected into the language model as an explicit intermediate interface, encouraging relationally consistent and safety-aware reasoning. To evaluate this problem, we present Falcon-X, a benchmark that unifies dense grounding with structured supervision over component completeness and risk inference in cluttered X-ray imagery. Experiments show that while existing multimodal models adapt to appearance, they struggle with compositional safety reasoning. Falcon improves functional grounding and produces more coherent threat assessments, establishing compositional safety reasoning as a distinct evaluation paradigm for multimodal systems.

Vision-Language Models (VLMs) are AI systems that process both images and text, yet they often struggle with compositional visual reasoning questions that require chaining multiple steps together, such as identifying objects, counting them, and comparing the results. Existing approaches improve this reasoning by training models on human-written step-by-step explanations, but creating these annotations is expensive and difficult to scale. We propose a self-questioning framework that trains a VLM to break visual questions into smaller sub-questions and answer each one before producing a final response, using a reinforcement learning algorithm called Group Relative Policy Optimization (GRPO). The model is never shown examples of how to decompose questions, it discovers this behavior on its own, guided by a reward signal that scores whether the output contains sub-questions and whether the final answer is correct. We apply this framework to a 3-billion-parameter model, training on both synthetic scenes of geometric shapes (CLEVR) and real-world photographs (A-OKVQA). On A-OKVQA, both self-questioning and standard reinforcement learning substantially improve accuracy over the untrained model (52.2% and 51.6% vs. 46.8%). We introduce the first self-questioning VLM by rewarding not only the final answer like standard RL but additionally for generating intermediate sub-questions, enabling it to discover compositional decomposition strategies. These results suggest that teaching AI systems to ask themselves intermediate questions is a promising strategy for complex visual reasoning, particularly when the difficulty of a question warrants explicit step-by-step decomposition.

Contrastively trained vision-language models like CLIP, have made remarkable progress in learning joint image-text representations, but still face challenges in compositional understanding. They often exhibit a "bag-of-words" behavior--struggling to capture the object relations, attribute-object bindings, and word order dependencies. This limitation arises not only from the reliance on global, single-vector representations for optimization, but also from the insufficient exploitation and modeling of the rich compositional information inherently present in paired image text data. In this work, we propose MACCO (MAsked Compositional Concept MOdeling), a framework that masks compositional concepts in one modality and reconstructs them conditioned on the full contextual information from the other, enabling the model to capture and align cross-modal compositional structures more effectively. To facilitate this process, we introduce two auxiliary objectives that jointly align and regularize masked features both inter-modally and intra-modally. Extensive experiments on five compositional benchmarks, along with in-depth analyses, demonstrate that our approach not only significantly enhances compositionality in VLMs but also improves their ability to capture syntactic structure and linguistic information. Additionally, the improved compositionality also benefits text-to-image generation and multimodal large language model. Code is available at https://github.com/hiker-lw/MACCO.

Vision-Language Models (VLMs) struggle with compositional reasoning that requires understanding inter-object relationships. A natural remedy is to inject explicit scene graph triplets $\langle s, p, o \rangle$ from an off-the-shelf scene graph generator (SGG), but we show this backfires: discrete text labels collide with the continuous visual modality, degrading GQA accuracy from 60.38% to 58.86%. We propose HyperVis, which bypasses the SGG semantic bottleneck entirely. From $N$ class-agnostic region proposals, we compute a dense $O(N^2)$ visual relation tensor via spatially-biased cross-attention, project it onto a Lorentz hyperboloid, and enforce hierarchy through spatial physics, namely IoA-driven entailment cones and exterior-angle repulsion. We discover that HyperVis contributes in two complementary ways: (1) as a training-time regularizer, the hyperbolic relational losses shape LoRA representations that improve generative VQA (GQA 61.03% vs.\ 57.21% for LoRA fine-tuning without relational losses, recovering and surpassing the baseline); and (2) as an inference-time relational encoder, hyperbolic prefix tokens boost discriminative compositional scoring (SugarCrepe 79.94%, $+$6.25pp over baseline). The learned curvature stabilises at $κ{=}4.0$, an order of magnitude above prior hyperbolic VLMs where $κ$ typically collapses toward zero, indicating that continuous visual features genuinely require the exponential volume of strongly curved space. A controlled Euclidean ablation confirms this decomposition: the relational pipeline regularises LoRA comparably in flat space (GQA 60.81%), but the compositionality gain is specifically hyperbolic (SugarCrepe $+$4.58pp over Euclidean), with entailment loss ${\sim}6{\times}$ higher in Euclidean training. Codes are available at TBA.

Chain-of-thought (CoT) reasoning has significantly improved the reasoning ability of large vision-language models (LVLMs) by verbalizing intermediate reasoning steps in natural language. However, such discrete textual rationales are often insufficient for encoding continuous visual evidence. Recent work addresses this limitation by moving reasoning into continuous latent space. Despite promising progress, existing methods leave latent reasoning insufficiently connected to the compositional and relational structure of visual evidence. To address this gap, we introduce ReGuLaR, a relation grounded latent reasoning framework that explicitly grounds latent states in these critical yet overlooked visual evidence. ReGuLaR uses a training-time ReGFormer to focus latent reasoning on question-relevant objects and inter-object relations, while at inference time the model reasons and generates answers without invoking the ReGFormer. To support training ReGuLaR, we construct RGROUNDING-351K, a real-world vision-language dataset annotated with key object bounding boxes and inter-object relations. Extensive experiments across diverse benchmarks show that ReGuLaR consistently outperforms existing approaches and achieves state-of-the-art performance. We include our code in the submission and will release the code and training data publicly upon acceptance.

Vision-Language Models demonstrate remarkable capabilities but often struggle with compositional reasoning, exhibiting vulnerabilities regarding word order and attribute binding. This limitation arises from a scarcity of informative samples needed to differentiate subtle semantic variations during contrastive pretraining. Although hard negative mining offers a promising remedy, existing methods lack explicit mechanisms to dictate which linguistic elements undergo modification. Instead of engineering generative architectures, this study establishes lexical concreteness as a fundamental determinant of negative sample efficacy. Modifying highly concrete terms generates more pronounced structural and visual discrepancies, providing a substantially stronger learning signal. Leveraging this principle, ConcretePlant is proposed to systematically isolate and manipulate perceptually grounded concepts. Analyses of the InfoNCE further reveals a severe gradient imbalance, where easily distinguishable pairs disproportionately overwhelm the optimization process and restrict the bandwidth available for nuanced learning. To resolve this degradation, the Cement loss is formulated utilizing a margin-based approach. By correlating psycholinguistic scores with sample difficulty, this objective dynamically calibrates the penalization applied to individual training pairs. Comprehensive evaluations substantiate these theoretical claims. The integrated framework, designated as Slipform, achieves state-of-the-art accuracy across diverse compositional evaluation benchmarks, general cross-modal retrieval, single and multi label linear probing.

Dual-encoder Vision-Language Models (VLMs) such as CLIP are often characterized as bag-of-words systems due to their poor performance on compositional benchmarks. We argue that this limitation may stem less from deficient representations than from the standard inference protocol based on global cosine similarity. First, through controlled diagnostic experiments, we show that explicitly enforcing fine-grained region-segment alignment at inference dramatically improves compositional performance without updating pretrained encoders. We then introduce a lightweight transformer that learns such alignments directly from frozen patch and token embeddings. Comparing against full fine-tuning and prior end-to-end compositional training methods, we find that although these approaches improve in-domain retrieval, their gains do not consistently transfer under distribution shift. In contrast, learning localized alignment over frozen representations matches full fine-tuning on in-domain retrieval while yielding substantial improvements on controlled out-of-domain compositional benchmarks. These results identify global embedding matching as a key bottleneck in dual-encoder VLMs and highlight the importance of alignment mechanisms for robust compositional generalization.

Accurate semantic understanding of complex traffic signs-including those with intricate layouts, multi-lingual text, and composite symbols-is critical for autonomous driving safety. Current models, both specialized small ones and large Vision Language Models (VLMs), suffer from a significant bottleneck: a lack of compositional generalization, leading to failure when encountering novel sign configurations. To overcome this, we propose SignReasoner, a novel paradigm that transforms general VLMs into expert traffic sign reasoners. Our core innovation is Functional Structure Unit (FSU), which shifts from common instance-based modeling to flexible function-based decomposition. By breaking down complex signs into minimal, core functional blocks (e.g., Direction, Notice, Lane), our model learns the underlying structural grammar, enabling robust generalization to unseen compositions. We define this decomposition as the FSU-Reasoning task and introduce a two-stage VLM post-training pipeline to maximize performance: Iterative Caption-FSU Distillation that enhances the model's accuracy in both FSU-reasoning and caption generation; FSU-GRPO that uses Tree Edit Distance (TED) to compute FSU differences as the rewards in GRPO algorithm, boosting reasoning abilities. Experiments on the newly proposed FSU-Reasoning benchmark, TrafficSignEval, show that SignReasoner achieves new SOTA with remarkable data efficiency and no architectural modification, significantly improving the traffic sign understanding in various VLMs.

Vision-language models (VLMs) excel at image-text retrieval yet persistently fail at compositional reasoning, distinguishing captions that share the same words but differ in relational structure. We present, a unified evaluation and augmentation framework benchmarking four architecturally diverse VLMs,CLIP, BLIP, LLaVA, and Qwen3-VL-8B-Thinking,on the Winoground benchmark under plain and scene-graph-augmented regimes. We introduce a dependency-based TextSceneGraphParser (spaCy) extracting subject-relation-object triples, and a Graph Asymmetry Scorer using optimal bipartite matching to inject structural relational priors. Caption ablation experiments (subject-object masking and swapping) reveal that Qwen3-VL-8B-Thinking achieves a group score of 62.75, far above all encoder-based models, while a proposed multi-turn SG filtering strategy further lifts it to 66.0, surpassing prior open-source state-of-the-art. We analyze the capability augmentation tradeoff and find that SG augmentation benefits already capable models while providing negligible or negative gains for weaker baselines. Code: https://github.com/amartyacodes/Inference-Time-Structural-Reasoning-for-Compositional-Vision-Language-Understanding

Contrastive vision-language (V&L) models remain a popular choice for various applications. However, several limitations have emerged, most notably the limited ability of V&L models to learn compositional representations. Prior methods often addressed this limitation by generating custom training data to obtain hard negative samples. Hard negatives have been shown to improve performance on compositionality tasks, but are often specific to a single benchmark, do not generalize, and can cause substantial degradation of basic V&L capabilities such as zero-shot or retrieval performance, rendering them impractical. In this work we follow a different approach. We identify two root causes that limit compositionality performance of V&Ls: 1) Long training captions do not require a compositional representation; and 2) The final global pooling in the text and image encoders lead to a complete loss of the necessary information to learn binding in the first place. As a remedy, we propose two simple solutions: 1) We obtain short concept centric caption parts using standard NLP software and align those with the image; and 2) We introduce a parameter-free cross-modal attention-pooling to obtain concept centric visual embeddings from the image encoder. With these two changes and simple auxiliary contrastive losses, we obtain SOTA performance on standard compositionality benchmarks, while maintaining or improving strong zero-shot and retrieval capabilities. This is achieved without increasing inference cost. We release the code for this work at https://github.com/saic-fi/concept_centric_clip.

The rapid compression of large vision-language models (VLMs) for edge deployment raises an underexplored question: do compact models fail differently, not merely more often? This study compares a 7-billion-parameter quantised VLM (Qwen2.5-VL-7B, 4-bit NF4) against a 500-million-parameter FP16 model (SmolVLM2-500M) across 4,000 samples from VQAv2 and COCO Captions. A three-category error taxonomy (Object Blindness, Semantic Drift, Prior Bias) is applied as a diagnostic framework. A text-only GPT-4o judge reveals Semantic Drift (B) as the dominant failure mode on VQAv2 and on COCO for Qwen, with a mixed Object Blindness / Semantic Drift profile for SmolVLM2 on COCO; Prior Bias (C) is present on VQAv2 but absent on COCO for both models. Confidence calibration is measured via Expected Calibration Error (ECE) using geometric mean token probability, compositional reasoning is probed with structured negation probes across four templates, and a blur robustness experiment completes the evaluation. For this model pair, the compact model exhibits a qualitatively distinct failure signature: a 12.5pp larger negation collapse (-33.2pp vs. -20.8pp, Wald 95% CI [8.2, 16.8]pp, p < 10^-8), driven almost entirely by COCO while the VQAv2 gap is not statistically significant (4.5pp, p=0.19). The most discriminating template is false_yn: SMOLVLM2-500M responds "Yes" (incorrectly claiming a depicted object is absent) on 100% of COCO trials vs. 14% for Q WEN 2.5-VL-7B. Asymmetric dataset-dependent miscalibration and a blur experiment with two controlled ablations complete the analysis. The fully reproducible pipeline is released for systematic safety auditing of compressed VLMs prior to edge deployment.

Existing visual privacy benchmarks largely treat privacy as a binary property, labeling images as private or non-private based on visible sensitive content. We argue that privacy is fundamentally compositional. Attributes that are benign in isolation may combine to produce severe privacy violations. We introduce the Compositional Privacy Risk Taxonomy (CPRT), a regulation-aware framework that organizes visual attributes according to standalone identifiability and compositional harm potential. CPRT defines four graded severity levels and is paired with an interpretable scoring function that assigns continuous privacy severity scores. We further construct a taxonomy-aligned dataset of 6.7K images and derive compositional risk scores. By evaluating frontier and open-weight VLMs we find that frontier models align well with compositional severity when provided structured guidance, but systematically underestimate composition-driven risks. Smaller models struggle to internalize graded privacy reasoning. To bridge this gap, we introduce a deployable 8B SFT model that closely matches frontier-level performance on compositional privacy assessment

Recent advancements extend Multimodal Large Language Models (MLLMs) beyond standard visual question answering to utilizing external tools for advanced visual tasks. Despite this progress, precisely executing and effectively composing diverse tools for complex tasks remain persistent bottleneck. Constrained by sparse tool-sets and simple tool-use trajectories, existing benchmarks fail to capture complex and diverse tool interactions, falling short in evaluating model performance under practical, real-world conditions. To bridge this gap, we introduce VisualToolChain-Bench(VTC-Bench), a comprehensive benchmark designed to evaluate tool-use proficiency in MLLMs. To align with realistic computer vision pipelines, our framework features 32 diverse OpenCV-based visual operations. This rich tool-set enables extensive combinations, allowing VTC-Bench to rigorously assess multi-tool composition and long-horizon, multi-step plan execution. For precise evaluation, we provide 680 curated problems structured across a nine-category cognitive hierarchy, each with ground-truth execution trajectories. Extensive experiments on 19 leading MLLMs reveal critical limitations in current models' visual agentic capabilities. Specifically, models struggle to adapt to diverse tool-sets and generalize to unseen operations, with the leading model Gemini-3.0-Pro only achieving 51% on our benchmark. Furthermore, multi-tool composition remains a persistent challenge. When facing complex tasks, models struggle to formulate efficient execution plans, relying heavily on a narrow, suboptimal subset of familiar functions rather than selecting the optimal tools. By identifying these fundamental challenges, VTC-Bench establishes a rigorous baseline to guide the development of more generalized visual agentic models.