Cambrian-S: Towards Spatial Supersensing in Video
This work addresses the need for more comprehensive spatial cognition in AI systems, moving beyond incremental improvements to propose a new foundational direction for multimodal intelligence.
The paper tackles the problem of advancing multimodal intelligence beyond linguistic understanding by introducing spatial supersensing, a paradigm that includes semantic perception, event cognition, spatial inference, and predictive modeling. It presents the VSI-SUPER benchmark and Cambrian-S model, achieving a +30% improvement on VSI-Bench, but shows that scale alone is insufficient, with a proof-of-concept predictive sensing approach outperforming proprietary baselines.
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.