After testing today, the setting does not seem to help, like at all. Opus 5.5 just keeps writing essays.

And it manages to overcomplicate even the most simple concepts in its explanations.

It's still very poor compared to OpenAI models.