Beyond Visual CoT: Internalized Visual Thinking for Proactive Video Reasoning

This post introduces Internalized Visual Thinking (IVT), a post-training framework that jointly optimizes textual prediction and next-embedding prediction over unlabeled videos. IVT predicts latent representations of future frames together with the target textual answer, encouraging the model to capture motion, object transitions, interactions, and latent intent. At inference, IVT generates the answer directly without synthesizing or re-encoding future frames. IVT improves over text-only post-training across all six evaluation settings while retaining the same efficient inference pathway. Compared with Visual CoT, IVT achieves comparable or better performance and reduces end-to-end latency by more than 5×.