Fireworks AI Production Inference Checklist
Choose Serverless or on-demand capacity for each workload, plan retries for 429s and 503s, and keep the balance and spend limit from pausing traffic. Then clear identity, retention and audit checks before production.
IncludesCapacity, limits, billing, model-launch and security checks, with the Fireworks page behind each
- Before workPick capacity per workload
- During workHandle 429s, 503s and pauses
- Before scalingReserve capacity for new models