Authors: Ruisi Wang, Zhongang Cai, Fanyi Pu, Junxiang Xu, Wanqi Yin, Maijunxian Wang, Ran Ji, Chenyang Gu, Bo Li, Ziqi Huang, Hokin Deng, Dahua Lin, Ziwei Liu, Lei Yang
Organizations: SenseTime Research · Nanyang Technological University · University of California, Berkeley · University of California, San Diego · Carnegie Mellon University
Recent advances in video generation have revealed an unexpected phenomenon: diffusion-based video models exhibit non-trivial reasoning capabilities. Prior work attributes this to a Chain-of-Frames (CoF) mechanism, where reasoning is assumed to unfold sequentially across video frames. In this work, we challenge this assumption and uncover a fundamentally different mechanism. We show that reasoning in video models instead primarily emerges along the diffusion denoising steps. Through qualitative analysis and targeted probing experiments, we find that models explore multiple candidate solutions in early denoising steps and progressively converge to a final answer, a process we term Chain-of-Steps (CoS). Beyond this core mechanism, we identify several emergent reasoning behaviors critical to model performance: (1) working memory that supports tasks requiring consistent reference, such as object permanence; (2) self-correction and enhancement, allowing recovery from incorrect intermediate solutions; and (3) perception before action, where early steps establish semantic grounding and later steps perform structured manipulation. Moreover, analysis of Diffusion Transformer layers shows that middle layers conduct key reasoning procedures. Motivated by these insights, we present a simple Training-Free Ensemble (TFE) as a proof-of-concept, demonstrating how reasoning can be improved by ensembling latent trajectories from identical models with different random seeds. Overall, our work provides the first systematic dissection of the mechanisms underlying video reasoning, offering a foundation to guide future research in better exploiting the inherent reasoning dynamics of video models as a new substrate for intelligence.
Figures & tables
Figure 1: Chain-of-Steps. We discover that video reasoning occurs along the diffusion steps with surprising emergent behaviors such as making multiple possible moves ( e.g. , paths) simultaneously at early steps, gradually pruning suboptimal choices during middle steps, and reaching a final decision at the late steps. This maze-solving example asks the model to start from the green circle in the top-left corner and find the red rectangle. Key regions of interest are color-coded and enlarged on the right.
Figure 2: Chain-of-Steps elicits reasoning along the diffusion process. We observe that video reasoning models explore multiple possible solutions simultaneously in the early denoising steps before converging to a final outcome in later steps. Specifically, we observe: (a) two potential routes (cyan arrows highlight the "imaginary traces") for the robot; (b) two possible placements of the "O" piece; (c) multiple candidate end positions for the plant; (d) simultaneous selection of two diamonds; (e) large and small circles overlapping with each other; and (f) all possible rotations of the L-shaped object superimposed. (g) Average number of alternative intermediate candidates observed at each diffusion step.
Figure 3: Noise perturbation and information flow. (a) Illustration of noise injection schemes; "Noise at Step" suffers more significant corruption than "Noise at Frame". (b) Performance drop with the two noise injection schemes. X-axis is the injection index (either diffusion step or frame). (c) Information flow across denoising steps (CKA dissimilarity: 1.0 indicates complete corruption, 0.0 indicates no effect).
Figure 4: Emergent reasoning behaviors: memory and self-correction. (a) The center point is retained to guide the return motion. (b) The contour of the occluded small teddy bear is preserved, enabling the model to address object permanence. (c) The trajectory of the ball gradually extends and becomes complete. (d) The missing cube only appears in the later diffusion steps. Cyan boxes are added for illustration; they are not part of the generated video.
Figure 5: Emergent reasoning behavior: understanding before reasoning. (a) Early diffusion steps identify the car as the object of interest with no motion, while later steps introduce action and simulate physical interactions. (b) Early steps recognize the door as the target object with no motion, and later steps manipulate it.
Figure 6: Layer specialization. (a) Layer-wise activation visualization shows that early layers of the video reasoning DiT tend to focus on background structures, whereas later layers carry out reasoning-related computations. (b) Layer-wise latent swapping reveals that certain middle layers ( e.g. , Layer 21 in this case) contain critical reasoning information that strongly influences the final outcome.
In-Domain by Category
Out-of-Domain by Category
Models
Overall
Avg.
Abst.
Know.
Perc.
Spat.
Trans.
Avg.
Abst.
Know.
Perc.
Spat.
Trans.
Human
0.974
0.960
0.919
0.956
1.00
0.95
1.00
0.988
1.00
1.00
0.990
1.00
0.970
Open-source Video Models
CogVideoX1.5-5B-I2V [ 73 ]
0.273
0.283
0.241
0.328
0.257
0.328
0.305
0.262
0.281
0.235
0.250
0.254
0.282
HunyuanVideo-I2V [ 29 ]
0.273
0.280
0.207
0.357
0.293
0.280
0.316
0.265
0.175
0.369
0.290
0.253
0.250
Wan2.2-I2V-A14B [ 60 ]
0.371
0.412
0.430
0.382
0.415
0.404
0.419
0.329
0.405
0.308
0.343
0.236
0.307
Table 1: Benchmarking results on VBVR-Bench. Overall In-Domain (ID) and Out-of-Domain (OOD) scores are reported alongside category-wise performance. Higher is better. Bold : best in group; underline : second best.
Method
VBVR- Bench
V-Reason Bench
MME- CoF
RISE- Video
VBVR-Wan2.2
0.69
8.94
1.30
61.60
+ TFE
0.72
12.12
1.52
65.85
VBVR-Wan2.1
0.59
12.66
0.36
37.35
+ TFE
0.61
13.35
0.38
42.92
VBVR-LTX2.3
0.52
3.61
0.12
32.75
+ TFE
0.53
3.19
0.27
35.85
Table 2: TFE results on four benchmarks
Appendix figures & tables18 assets
Supplementary material from the paper’s appendix.
Appendix
# Seeds
Overall
ID
OOD
1
0.685
0.760
0.610
2
0.688
0.770
0.606
3 (TFE default)
0.716
0.780
0.650
4
0.729
0.781
0.676
5
0.720
0.782
0.658
6
0.675
0.771
0.579
Appendix
Table 3: TFE scaling curve
In-Domain by Category
Out-of-Domain by Category
Aggregated Layers
Overall
Avg.
Abst.
Know.
Perc.
Spat.
Trans.
Avg.
Abst.
Know.
Perc.
Spat.
Trans.
baseline(VBVR-Wan2.2) [ 61 ]
0.685
0.760
0.724
0.750
0.782
0.745
0.833
0.610
0.768
0.572
0.547
0.618
0.615
0-9
0.688
0.774
0.754
0.741
0.808
0.746
0.835
0.602
0.805
0.635
0.485
0.597
0.642
0-39
0.690
0.767
0.733
0.737
0.807
0.747
0.825
0.613
0.830
0.606
0.482
0.630
0.657
20-29(Training-Free Ensemble)
0.716
0.780
0.760
0.744
0.809
0.749
0.858
0.650
0.803
0.705
0.531
0.639
0.716
Appendix
Table 4: Comparison of VBVR-Bench performance across different layer windows at diffusion step s=0 . Mid-layer aggregation (20–29) achieves the best overall performance (0.716) by capturing the critical reasoning-active window. Bold : best in group; underline : second best.
Figure 7: The outcome-flip rates of layer-wise swapping experiments always peak at middle layers across different seeds
Figure 8: Layer-wise token activation visualization across all 40 DiT blocks. Rows correspond to layers 0–39 (from top to bottom), while columns represent video frames.
Figure 9: (Continued) Layer-wise token activation visualization across all 40 DiT blocks. Rows correspond to layers 0–39 (from top to bottom), while columns represent video frames.
Configuration
Overall
In-Domain
Out-of-Domain
Chronoedit
0.581
0.637
0.524
5 frames
0.619
0.688
0.549
9 frames
0.632
0.716
0.549
17 frames
0.663
0.743
0.582
33 frames
0.685
0.685
0.685
65 frames
0.675
0.760
0.591
Appendix
Table 5: Comparison of Model Performance Across Frame Counts. Chronoedit here could be considered as an single-frame version of VBVR-Wan2.2. The original VBVR-Wan2.2 operates on ∼ 100 frames on average.
Figure 15
Figure 10: Qualitative visualizations of the distilled model.
Figure 11: Qualitative examples on VBVR-LTX2.3.
Figure 12: (Continued) Qualitative examples on VBVR-LTX2.3.
Figure 13: Qualitative examples on VBVR-Wan2.1.
Figure 14: (Continued) Qualitative examples on VBVR-Wan2.1.
Figure 15: More visualizations of "Multi-Path Exploration" phenomenon in Sec. 3.1.1 .
Figure 16: (Continued) More visualizations of "Multi-Path Exploration" phenomenon in Sec. 3.1.1 .
Figure 17: More visualizations of "Superposition-based Exploration" phenomenon in Sec. 3.1.2 .
Figure 18: More visualizations of "Memory" phenomenon in Sec. 4.1 .
Figure 19: More visualizations of "Self-correction and Enhancement" phenomenon in Sec. 4.2 .
Figure 20: More visualizations of "Self-correction and Enhancement" phenomenon in Sec. 4.2 .