MMCL-Bench: Multimodal Context Learning from Visual Rules, Procedures, and Evidence
This benchmark identifies and measures a previously unaddressed capability gap in multimodal AI systems, providing a structured evaluation for an important unsolved problem.
MMCL-Bench introduces a benchmark for multimodal context learning, requiring models to learn rules, procedures, and patterns from visual contexts and apply them to new instances. The strongest model solves fewer than one-third of tasks under strict evaluation, revealing a key capability bottleneck.
We introduce MMCL-Bench, a benchmark for multimodal context learning: learning task-local rules, procedures, and empirical patterns from visual or mixed-modality teaching context and applying them to new visual instances. Unlike text-only context learning or standard multimodal question answering, this setting requires models to recover and localize relevant evidence from images, screenshots, manuals, videos, and frame sequences before they can reason over the learned context. MMCL-Bench contains 102 tasks spanning three categories: rule system application, procedural task execution, and empirical discovery and induction. We evaluate frontier multimodal models with strict rubric-based scoring and find that current systems remain far from robust multimodal context learning, with even the strongest model solving fewer than one-third of tasks under strict evaluation. Diagnostic ablations and error analysis show that failures arise throughout the context-to-answer pipeline, including context anchoring, visual evidence extraction, context reasoning, and response construction. MMCL-Bench thus highlights multimodal context learning as an important unsolved capability bottleneck for current multimodal models.