How do you handle LLM provider outages?

Question

Accepted Answer

Multi-provider: Primary OpenAI → fallback Anthropic → fallback cached response.