Cambrian-S: Towards Spatial Supersensing in Video

TL;DR

This paper introduces the concept of spatial supersensing in video understanding, proposes new benchmarks to evaluate it, and demonstrates that scaling data alone is insufficient, advocating for predictive sensing to advance true multimodal intelligence.

Contribution

It defines the paradigm of spatial supersensing, introduces VSI-SUPER benchmarks, and presents a predictive sensing approach that significantly improves spatial understanding in videos.

Findings

VSI-SUPER benchmarks reveal current models' limitations in spatial cognition.

Scaling data improves performance but does not fully achieve spatial supersensing.

Predictive sensing with surprise-driven memory outperforms existing baselines.

Abstract

We argue that progress in true multimodal intelligence calls for a shift from reactive, task-driven systems and brute-force long context towards a broader paradigm of supersensing. We frame spatial supersensing as four stages beyond linguistic-only understanding: semantic perception (naming what is seen), streaming event cognition (maintaining memory across continuous experiences), implicit 3D spatial cognition (inferring the world behind pixels), and predictive world modeling (creating internal models that filter and organize information). Current benchmarks largely test only the early stages, offering narrow coverage of spatial cognition and rarely challenging models in ways that require true world modeling. To drive progress in spatial supersensing, we present VSI-SUPER, a two-part benchmark: VSR (long-horizon visual spatial recall) and VSC (continual visual spatial counting). These tasks require arbitrarily long video inputs yet are resistant to brute-force context expansion. We then test data scaling limits by curating VSI-590K and training Cambrian-S, achieving +30% absolute improvement on VSI-Bench without sacrificing general capabilities. Yet performance on VSI-SUPER remains limited, indicating that scale alone is insufficient for spatial supersensing. We propose predictive sensing as a path forward, presenting a proof-of-concept in which a self-supervised next-latent-frame predictor leverages surprise (prediction error) to drive memory and event segmentation. On VSI-SUPER, this approach substantially outperforms leading proprietary baselines, showing that spatial supersensing requires models that not only see but also anticipate, select, and organize experience.