Applied AI·Production patterns and cost
the provider had an incident and your feature stayed up, because a second model was configured and had actually been tested.
Multi-provider failover
Draft summary, pending review
Keeping a second provider or model configured and tested for when the primary degrades or goes down, with prompts portable enough to survive the swap. Providers have incidents like anyone else; your uptime should not inherit them.