Two-minute worlds · Generated worlds
What happens if we keep going?
Two minutes in a generated world, with plenty of turns along the way.
Clip files & individual measurements
Matrix-Game 3 Distilled
Skywork
- AI landmark retention
- 29.2 %
- AI landmark contradiction
- 4.2 %
- AI landmark uncertainty
- 66.7 %
- Return judgments completed
- 100.0 %
- Judge first-attempt coverage
- 100.0 %
- Turn pairs tracked
- 100.0 %
- Rotation pairs qualified
- 100.0 %
- Commanded warp discrepancy
- 0.757 % diagonal
- Qualified rotation error
- 0.927 degrees
- Qualified yaw direction
- 99.4 %
- Idle warp discrepancy
- 0.018 % diagonal
- First decoded worker chunk
- 6.1 seconds
- Steady worker throughput
- 5.115 frames/s
- Peak allocated GPU memory
- 31.179 GiB
- Session process time
- 445.3 seconds
- Late / early worker latency
- 1.011 ratio
Matrix-Game 3 Base
Skywork
- AI landmark retention
- 83.3 %
- AI landmark contradiction
- 12.5 %
- AI landmark uncertainty
- 4.2 %
- Return judgments completed
- 100.0 %
- Judge first-attempt coverage
- 100.0 %
- Turn pairs tracked
- 85.8 %
- Rotation pairs qualified
- 85.8 %
- Commanded warp discrepancy
- 1.180 % diagonal
- Qualified rotation error
- 1.599 degrees
- Qualified yaw direction
- 94.0 %
- Idle warp discrepancy
- 0.019 % diagonal
- First decoded worker chunk
- 164.5 seconds
- Steady worker throughput
- 0.235 frames/s
- Peak allocated GPU memory
- 31.196 GiB
- Session process time
- 8383.7 seconds
- Late / early worker latency
- 1.000 ratio
Each clip shows one registered input and seed. The full study includes every seed and failure.
Look for thisCompare the early returns with the late ones. Watch for objects shifting, disappearing, or becoming hard to recognize.
Every run, including failures (data file) (opens in a new tab)
Choose a scene or change the models
Examples stay in their registered order. Every seed and any failed runs remain in the full results.
Compare the numbers across every run
Matrix-Game 3 Distilled − Matrix-Game 3 Base. Uses every jointly scored case and seed for these two configurations, across the full study.
| Measurement | Left − right | 95% interval | Paired coverage |
|---|---|---|---|
| AI landmark retentionpercentage points | -35.4 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| AI landmark contradictionpercentage points | -12.5 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| AI landmark uncertaintypercentage points | +47.9 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Return judgments completedpercentage points | 0.0 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Judge first-attempt coveragepercentage points | 0.0 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Turn pairs trackedpercentage points | +7.1 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Rotation pairs qualifiedpercentage points | +7.1 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Commanded warp discrepancy% diagonal | -0.326 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Qualified rotation errordegrees | -0.619 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Qualified yaw directionpercentage points | +5.8 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Idle warp discrepancy% diagonal | -0.003 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| First decoded worker chunkseconds | -158.9 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Steady worker throughputframes/s | +4.881 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Peak allocated GPU memoryGiB | -0.017 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Session process timeseconds | -7928.6 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
| Late / early worker latencyratio | +0.011 | Interval unavailable | 2 matched pairs2 groups · 0 excluded pairs |
A positive difference means the left model has a higher value; a negative difference means a lower value. Whether that is better depends on the metric. Excluded pairs include inapplicable or unscored measurements. These differences use matched runs and can differ from subtracting the displayed averages. Intervals that cross zero do not establish a clear direction of difference.
This small stress test covers just two scenes and one random seed. Its AI landmark labels describe these clips, not a general winner.
Results corrected on 2026-10-16
Preserve original raw videos and worker snapshots. For SANA score registered camera timestamp f using saved output frame f-1; evaluate interframe motion against camera poses[1:1+n_output]. MG3 retains frame f and poses[0:n_output]. This source-defined alignment correction preceded all primary scoring; original outputs are retained.
Analysis and original results (opens in a new tab)How we tested it — results, method & every run
Two-minute world stress sessions
Which released configuration preserves recognizable scene landmarks across repeated one- and two-minute turn-away/return sessions, follows camera actions, and sustains generation throughput?
2 cases · 1 seed · 3 compared configurations · completed 2026-10-16 · descriptive stress subset
What we found. Two scenes, one seed each: this stress subset describes these outputs only and has no confidence intervals. Base received 83.3% retained-landmark labels at 0.236 worker fps; Distilled received 47.9% at 5.117 fps, with 50.0% unclear labels. These scores combine scene visibility and persistence; they do not isolate memory. Landmark scores are AI estimates, not validated game-state memory. Compare MG3 Base and Distilled within their shared next-action interface; SANA receives a future camera path and caption. Worker speed excludes browser delivery.
Watch the turn-away/return cycles and compare the starting scene with later returns. Scores concern three declared visible landmarks, not every object. SANA has extra future-camera and caption inputs; cross-group clips are illustrative, without paired rankings.
| Model configuration | AI landmark retention% · higher is better | AI landmark contradiction% · lower is better | AI landmark uncertainty% · lower is better | Return judgments completed% · higher is better | Judge first-attempt coverage% · higher is better | Turn pairs tracked% · higher is better | Rotation pairs qualified% · higher is better | Commanded warp discrepancy% diagonal · lower is better | Qualified rotation errordegrees · lower is better | Qualified yaw direction% · higher is better | Idle warp discrepancy% diagonal · lower is better | First decoded worker chunkseconds · lower is better | Steady worker throughputframes/s · higher is better | Peak allocated GPU memory |
|---|
