Your primary LLM provider is fully down for an hour, and no tested fallback exists for your agent feature. The product also has an FAQ answer feature backed by a semantic cache. What is the best degradation mode?

1. A Queue all requests indefinitely and replay them once the primary provider recovers.
2. B Retry each failed call in a tight loop until the primary provider starts responding.
3. C Serve cached or reduced answers where possible and show a degraded-mode notice.
4. D Send all traffic to whichever provider is up and assume the prompts behave the same.