World Modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
47.7Exact Match (EM)
30
Jun 17, 2026
65.7Accuracy
20
Feb 26, 2026
52.8Matter Score
20
Feb 26, 2026
98.4Smelt
20
Feb 26, 2026
100Search
20
Feb 26, 2026
67.36Image Quality
15
May 8, 2026
0.3CD (Inner)
12
May 11, 2026
10.31FID
10
Jul 7, 2026
0.13CD Inner
10
May 11, 2026
2.36Instruction Following Score
8
Jun 16, 2026
59.7EWMScore
7
May 20, 2026
71.39Dynamic Score
7
Apr 10, 2026
41.31R Score
6
May 15, 2026
4.5R (Trajectory Fidelity)
6
May 15, 2026
14.8PSNR
6
May 15, 2026
14.59PSNR
6
May 15, 2026
100Next-Token Test Accuracy
5
May 25, 2026
18.7Rank @ 1
5
Apr 9, 2026
421.9FVD
4
Jul 7, 2026
392.2FVD
4
Jul 7, 2026
373.1FVD
4
Jul 7, 2026
23.48Exterior FID
4
Jun 12, 2026
9.51Exterior FID
4
Jun 12, 2026
26.711PSNR (10 frames)
4
May 19, 2026
73.163D Consistency
4
Mar 4, 2026