ResearchOct 27, 2026, 00:00 UTC

RynnVLA-Latent scales robot manipulation models with 3D-grounded latent actions

The 4B policy reports 98.8% success on LIBERO, 84.7% on LIBERO-Plus, and 88.6% average on the RoboTwin 2.0 bimanual benchmark, trained on 30.1K hours of human, robot, and simulation videos.

ModelRynnVLA-Latent
ModelRynnLAM
Topicresearch
Topicrobotics
Topicbenchmark

1outlet ran itRead at Hugging Face Papers

Status: Research
Published: Oct 27, 2026, 00:00 UTC That day
Outlets: 1
Written by: Claude How

Every outlet that ran it 1 article

Oct 27, 2026, 00:00 UTC
Hugging Face PapersResearchLead
RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions

All news