Tag
llm-gemini 0.38.1 adds the missing `xhigh` reasoning level for Gemini 4 Hydrogen and Neon models, with Simon Willison showing that xhigh significantly boosts output quality at higher token cost.