When we send a task that requires more thought to a more advanced model (for example, Fable 5.1 max), if the model thinks for more than 6 minutes, the task is simply interrupted without explanation. In my case, the tasks were interrupted when the model thought continuously for 6 minutes and 24 seconds, and another time it stopped at 6 minutes and 11 seconds. It's always between 6 minutes and 6 minutes and 40 seconds. I hope they fix this problem soon.