← Leaderboard

LaWAM

World-action
Tsinghua University · Jun 2026

Latent world-action model that predicts DINOv3 latent states with a 230M world model instead of generating video; from the RLinf project with Tsinghua as the first-listed affiliation.

Type
World-action
Parameters
2.3B
Backbone
Qwen3-VL-2B + DINOv3
Weights
Open
License
–
Commercial use
Unclear
Released
Jun 2026
Robots
Franka Panda (sim), Aloha-AgileX (sim), SO-101 (sim)
SO-101
–
LeRobot
–

Benchmark results

2 of 13 benchmarks
LIBEROSaturatedSimulation
#13 of 4698.6%, self-reported by the authors, not a leaderboard entry
SO-101 BenchLow-cost arm
#8 of 1139.9%, self-reported by the authors, not a leaderboard entry
Not reported on 11 benchmarks

Adoption

updated Nov 17, 2026
HF downloads · 30d
173
HF downloads · all time
488
HF likes
0
Wrong or missing number? Report an error