← Leaderboard

Being-H0.7

VLA
BeingBeyond · Apr 14, 2026

BeingBeyond’s 3B VLA, the successor to Being-H0.5 (InternVL3.5 understanding expert, Qwen3 action expert). Code and checkpoints are announced as coming soon.

Type
VLA
Parameters
3B
Backbone
InternVL3.5 + Qwen3
Weights
Unknown
License
Apache-2.0
Commercial use
Unclear
Released
Apr 14, 2026
Robots
Franka Panda (sim), Aloha-AgileX (sim), PND Adam-U, Unitree G1, Franka FR3
SO-101
–
LeRobot
–

Benchmark results

4 of 13 benchmarks
#14 of 2482.1%, self-reported by the authors, not a leaderboard entry
RoboTwin 2.0Near saturationSimulation
  • 89.9%, self-reported by the authors, not a leaderboard entryPaper / model card ↗May 2026· self-reported· non-standard
89.9%, self-reported by the authors, not a leaderboard entrynot ranked
CALVIN ABC→DNear saturationSimulation
#7 of 114.48 / 5, self-reported by the authors, not a leaderboard entry
LIBEROSaturatedSimulation
#6 of 4699.2%, self-reported by the authors, not a leaderboard entry
Not reported on 9 benchmarks
Robustness drop: 99.2% on LIBERO → 82.1% on LIBERO-Plus. The perturbed variants show how much of the LIBERO score is memorization.

Adoption

updated Nov 17, 2026
GitHub stars
1.2K
GitHub forks
60
Wrong or missing number? Report an error