TriViewBench: Controlled Complexity Scaling for Multi-View Structural Reasoning in MLLMs
For researchers evaluating MLLMs, this benchmark exposes fundamental scalability limitations in multi-view structural reasoning, showing that current models fail under controlled complexity.
TriViewBench is a controlled three-view visual reasoning benchmark for MLLMs, revealing that all 18 tested models show a consistent capability hierarchy (Local Decision > Object Counting > Global Recovery) with performance degrading monotonically as complexity increases, including a 59.14% drop in Object Counting and 80.02% collapse in Global Recovery. Chain-of-Thought prompting provides near-zero benefit (-0.16%), indicating the bottleneck is cross-view spatial representation.
Multimodal Large Language Models (MLLMs) demonstrate strong performance on standard visual question answering benchmarks, yet their scalability under controlled structural complexity remains poorly understood. We introduce TriViewBench, a controlled three-view visual reasoning benchmark constructed from synthetic 3D scenes with explicitly parameterized object count and occlusion. The benchmark contains 1,923 scenes and over 14K Question-Answer (QA) pairs organized into four complexity levels and three reasoning categories: Local Decision, Object Counting, and Global Recovery. We evaluate 18 open- and closed-source MLLMs under a unified prompting protocol. All 18 models exhibit an identical capability hierarchy without exception (Local Decision > Object Counting > Global Recovery), and performance degrades monotonically with complexity: Local Decision tasks decline modestly (12.11% relative drop), while Object Counting degrades substantially (59.14%) and Global Recovery collapses severely (80.02%). Error analysis on Object Counting reveals two mechanistically independent failure modes: single-view tasks are dominated by undercounting due to occlusion blindness, whereas the multi-view task reverses to overcounting due to cross-view identity confusion. Chain-of-Thought (CoT) prompting yields near-zero overall benefit ($Δ= -0.16\%$) and its effect on Global Recovery is strongly capability-gated, suggesting that the bottleneck lies in cross-view spatial representation rather than reasoning strategy. These findings reveal fundamental scalability limitations in current MLLMs and position TriViewBench as a controlled diagnostic framework for analyzing structural reasoning failures.