SL-AI drops GRaPE 2 reasoning models

GRaPE 2's focus on granular reasoning control positions it as a practical tool for local agentic workflows where efficiency and task-specific logic are paramount. The six thinking modes solve the overthinking problem in reasoning-heavy models, while direct training on code and browser agent examples suggests superior performance in autonomous task execution. At 5B and 9B parameters, these models are well-suited for high-performance local inference on consumer hardware.