A research analysis argues that multimodal large language models' multi-step reasoning failures, such as losing track of which objects have already been counted, stem from a "reference gap" rather than perception errors, and examines what visual scaffolds can and cannot fix. The work attributes these failures to models' inability to reliably track references across reasoning steps even when they recognize the visual content.
Multimodal large language models can recognize visual content yet struggle to reason reliably over it across multiple steps. In counting, a model may see all objects but lose track of which ones it has counted. We argue that such failures, commonly attributed to perception, reveal a reference gap: t
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.