MindJourney: Test-Time Scaling with World Models for Spatial Reasoning

Yilun Du (Google Deepmind / Harvard) · Chuang Gan (IBM) · Zheyuan Zhang (University of Notre Dame) · Jianwei Yang (Microsoft) · Reuben Tan (Boston University) · Siyuan Zhou (HKUST) · Yuncong Yang (University of Massachusetts at Amherst) · Jiageng Liu (University of Massachusetts at Amherst)

3d dynamicscamera trajectorycontrollable world modelegocentric motioninteractive explorationinternal modelmulti-view evidencereinforcement learningrobust 3d reasoningspatial reasoningspatial reasoning benchmarkspatialnavigatortest-time inferencetest-time scalingvideo diffusionvision-language models

Spatial reasoning in 3D space is central to human cognition and indispensable for embodied tasks such as navigation and manipulation. However, state-of-the-art vision–language models (VLMs) struggle frequently with tasks as simple as anticipating how a scene will look after an egocentric motion: they perceive 2D images but lack an internal model of 3D dynamics. We therefore propose SpatialNavigator, a test-time scaling framework that grants a VLM with this missing capability by coupling it to a controllable world model based on video diffusion. The VLM iteratively sketches a concise camera trajectory, while the world model synthesizes the corresponding view at each step. The VLM then reasons over this multi-view evidence gathered during the interactive exploration. Without any fine-tuning, our SpatialNavigator achieves an average 7.7\% performance boost on the representative spatial reasoning benchmark SAT, showing that pairing VLMs with world models for test-time scaling offers a simple, plug-and-play route to robust 3D reasoning. Meanwhile, our method also improves upon the test-time inference VLMs trained through reinforcement learning, which demonstrates the potential of our method that utilizes world models for test-time scaling.