An application receives repeated rate-limit responses after adding concurrent clients. What should the engineer investigate first?

1. Disable all delays and retry indefinitely
2. Whether responses use enough adjectives
3. Deployment quotas, request and token demand, and controlled retry behavior
4. Whether every caller has a different application icon