It that true?

> For Claude Sonnet 4.5, and GPT-5.1 we default to reporting high reasoning results, but when reported results are not available we use best available reasoning results.

https://storage.googleapis.com/deepmind-media/gemini/gemini_...