Manifold AI

WorldScape-R²SI↗

WorldScape-R²SI is a WAM from Manifold AI, released October 18, 2026. It has no overall rank on the Robot Intelligence Index, which needs both a simulation and a real-robot result. Its best result is #10 of 72 on RoboDojo Sim. Two-layer recursive self-improvement agent that revises skills around Manifold AI's WorldScape Policy world action model.

Release date: Oct 18, 2026

Developer
Manifold AI
Type
WAM
Model weights
Unknown
License
Not stated
Parameters
—
Pricing
Not published
Benchmarks
1
Overall rank
Not ranked
Needs a real-robot result to rank overall.
Real robots rank
Not ranked
No real-robot results yet.
Simulation rank
Not ranked
Needs two simulation benchmarks; has one (RoboDojo Sim).
Rating
1695
10601989
Cost / task · est.
$0.022
$0.0087$17.74
Time / task · est.
24s
9.5 s1 h 22 min
Rank on each benchmark
BenchmarkScoreRank
RoboDojo SimOverall46.610 / 72

Per task attempt; the tiles show the median across benchmarks.

Inference

Agent loop calling WorldScape Policy one action chunk at a time; no latency reported.