Which brain, when?

Hard problems that need step-by-step reasoning, agentic tool use, or visual inputs. The default for critical decisions.

High-volume, everyday chat where quality-per-dollar matters more than the absolute best answer.

Any task where prose quality, tone, or thoughtful analysis matters. Best writer in the catalog.

When you want Claude's writing feel at Haiku's speed and price — great for customer replies or bulk drafting.

The absolute lowest-cost brain in the catalog. Perfect for bulk summarisation over long docs.

When your input is a giant PDF, a full codebase, or hours of transcripts. Nothing else swallows 2M tokens.

Numeric, quantitative, or code-heavy tasks. Punches far above its price on STEM benchmarks.

Brainstorming, taglines, and multilingual copy. Say more with fewer tokens.

Storytelling, narrative marketing, and long creative briefs. Reads and remembers a lot.

General-purpose Meta open-weight model when you want a friendly voice with a large window.

Everyday chat with an open-weight lineage. A dependable, low-cost helper.

When data sovereignty or a European lineage matters, or you want a fast tool-using vision model outside the US Big Three.