Mini toolsXBai o4
By MetaStone AI
The project describes a “reflective generative” training form that combines long chain-of-thought reinforcement learning with process reward learning, aiming to improve deep reasoning and trajectory selection, with a dated “We release XBai o4” entry in the project news.
Text
Overview
XBai-o4 is an open large language model focused on test-time scaling and reflective reasoning improvements.
Last updated: April 6, 2026
