Next Forcing: Causal World Modeling with Multi-Chunk Prediction

Summary

Next Forcing is a multi-chunk prediction (MCP) framework for causal world modeling, presented by Gangwei Xu and colleagues in arXiv paper 2606.11187. Autoregressive video generation has become a powerful paradigm for World Action Models (WAMs), but existing approaches suffer from slow training convergence, limited converged accuracy at high frame rates (because supervision is confined to the current chunk without explicit signals about future dynamics), and slow inference due to iterative video denoising. Inspired by multi-token prediction in large language models, Next Forcing augments the main model with lightweight auxiliary MCP modules that simultaneously denoise video chunks across multiple future time scales. These modules form a causal chain across prediction depths, leveraging intermediate features from the main model's multi-layer fusion to predict future dynamics. According to the forum summary, the method achieves state-of-the-art results on RoboTwin (94.1/93.5%) with 2x inference speedup. Released June 9, 2026, in the computer vision field.