Anthropic makes Claude Opus 5.5 default with new prompting guidelines

The model automatically scales reasoning depth based on task complexity and effort level, with Medium effort matching prior High performance. It handles multi-step tasks more efficiently by self-managing interactions and directly producing final outputs. Image reading and document generation are significantly improved, especially at low effort levels.

Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.