T-pro 2.0: An Efficient Russian Hybrid-Reasoning Model and Playground Paper β’ 2512.10430 β’ Published 17 days ago β’ 112