Different tasks map to different capability tiers and are billed by tokens. You don’t pick models manually.
Fuxi automatically selects the right tier by task complexity—balancing quality and cost.
Auto-matched tier
Lightweight
Faster responses for completion, quick Q&A, and everyday coding help.
Selected automatically for light tasks
Auto-matched tier
Standard
Balanced capability for regular development, refactors, and docs.
Selected automatically for common work
Auto-matched tier
High performance
Stronger reasoning for complex changes, multi-file work, and long tasks.
Selected automatically for complex tasks
Exact rates are available on the usage page after login.