Stanford University

PhAI Labs, with The Chinese University of Hong Kong, University of Oxford, Stanford University, and Princeton University, has released JEPA-Anything, a cross-domain world model framework. Its orthogonal predictive decomposition (OPF) splits the target state into complementary subspaces, with separate prediction branches handling each. The team evaluated the model on visual, biological, clinical, control, molecular, physical-field, and weather systems. They report it outperforms standard JEPA on multiple prediction tasks and that internal factors yield scientific clues for cancer-cell intervention and analysis of planetary orbit regularities.

MetaRSI-v1, a meta recursive self-improvement architecture, has been released by CosmosMind jointly with Tsinghua University, Peking University, Stanford University, UC Berkeley, MIT and other institutions. It unifies what the team calls model-level, data-level and harness-level recursion, aiming to improve the process of self-improvement rather than a single model.

With no external teacher model, a 3B active-parameter model (Qwen3.5-35B-A3B) averaged a 10.9-point improvement across Terminal-Bench 2.1, SWE-bench Pro, a GPQA-Diamond subset and AIME; six frontier flagship models averaged a 7.3-point gain on Terminal-Bench 2.1. The RSI-Harness is open-sourced, and the collaboration plans to extend the loop to programmable instruments and automated labs.