Fable 5 chain-of-thought output appears in web interface

Users observed raw chain-of-thought reasoning logs from Fable 5, which appear as dense, non-standard text. This output is a byproduct of RLVR training, where reinforcement learning optimizes the model's internal reasoning process beyond plain English.