Scaling Video Generation for Reasoning: At What Cost?

Our controlled benchmark requires predicting nine prescribed moves of an initially solved 2x2x2 Rubik's Cube from a fixed view of three faces. It comes from a paper posted to arXiv on 30 September 2026. We study whether scaling video generation enables models to reason about hidden information from the past frames, and at what computational cost. Correct predictions require inferring how actions change hidden states, and the simulator provides exact ground truth for evaluation. Models learn plausible cube geometry early, while correct sticker configurations require substantially more training. Although validation MSE follows approximate power-law scaling, lower MSE loss does not reliably indicate downstream reasoning capabilities. Smaller autoregressive models achieve higher state accuracy with limited compute, while larger models reach higher accuracy after more training. At roughly 0.1 PF-days, the 70M-parameter model correctly predicts the visible sticker configuration in 44.6% of post-action frames, compared with 0.3% for the 1B model, which reaches 83.7% at 3.14 PF-days. Symbolic state supervision raises the 20M model's frame accuracy from 31.1% to 67.3% at the same training-data budget, suggesting that learning representations of state changes can complement scaling.

Why it counts

Our controlled benchmark requires predicting nine prescribed moves of an initially solved 2x2x2 Rubik's Cube from a fixed view of three faces.

Sources

Primary source: primary source

What is not known

This brief does not claim independent replication. Claims that appear only on X and not in the primary source stay unknown.