LinkMoo Jin Kim, Karl Pertsch, Siddharth Karamcheti et al.Research
Most competitive generalist robotics models have historically been proprietary or hard to adapt. OpenVLA releases model checkpoints and the PyTorch training pipeline on HuggingFace, providing an accessible foundation for research and custom multi-robot deployments.
The project demonstrates that a 7B parameter open architecture can outperform much larger proprietary systems (like 55B RT-2-X) on general manipulation by using diverse datasets and modern VLM backbones. Although Diffusion Policy models remain superior on narrow, high-precision single tasks, OpenVLA excels in multi-object scenes requiring language grounding and can be adapted efficiently using parameter-efficient fine-tuning.
Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.
Similar in content, found by the AI search.
Article:Moo Jin Kim et al.
Researchers from Stanford, UC Berkeley, Google DeepMind, and partner institutions introduce OpenVLA, an open-source 7-billion-parameter vision-language-action model. Trained on 970k real-world robot demonstrations, it surpasses closed models such as RT-2-X while enabling parameter-efficient fine-tuning on consumer-grade hardware.
Link:Zhe Liu, Jinghua Hou, Yuxiang Lu, et al.
Researchers from The University of Hong Kong and ACE Robotics introduce StreamPI, a framework that equips single-frame Vision-Language-Action (VLA) models such as π0.5 with temporal reasoning without adding new parameters. By bundling images and instructions into atomic temporal units and using random-interval training, StreamPI improves precision, memory, and robustness in real-robot manipulation.
Article:Peiyan Li, Yuze Zhu, Yixiang Chen, Qisen Ma, Yuan Xu, Jiabing Yang, He Guan, Yan Huang, Hongtao Wu, Xiao Ma, Tao Kong, Liang Wang, Tieniu Tan
Researchers introduce BridgeVLA++, an upgraded vision-language-action framework designed for 3D robotic manipulation. It introduces a unified spatio-temporal memory architecture into pre-trained vision-language models, enabling persistent spatial reasoning and temporal history tracking without losing data efficiency.
Article:Saeed Babamohamadi
NVIDIA details Cosmos 3, an open omni-model foundation that replaces conventional vision-language backbones with video world models to give robot policies predictive physical priors.
Link:Dyna Robotics
Dyna Robotics has launched DYNA-2, a World-Action Model (WAM) pre-trained on more than one million hours of egocentric human video. The model circumvents traditional teleoperation bottlenecks, transferring physical reasoning and action policies directly to diverse robotic hardware.
Link:Qwen
The Qwen team has released Qwen3.8-27B, a compact, dense vision-language model under the Apache-2.0 licence. The 27-billion-parameter model pairs Gated DeltaNet with conventional gated attention, handles images and videos natively, and introduces flexible reasoning controls.
Parts of this content were created with the assistance of AI.
