Data retention

Live
Separate request records, provider processing, and customer-owned output storage.
Open in ChatGPT
(opens in new tab)Last verified: 2026-08-11

Separate the layers

Request metadata, model input/output processing, and durable copies of generated assets have different lifecycles. Store only the application data you need and keep generated assets in customer-controlled storage when long-term access is required.

Published policies

Read the current Data Retention Policy, Data Deletion Policy, and Privacy Policy before production use. Model/provider differences remain relevant at the model layer.

Minimize sensitive data

Do not place secrets or unnecessary personal data in prompts, filenames, metadata, or logs. Apply your own deletion schedule to request IDs stored in your systems.
PreviousErrors, retries, and limits
NextCoding agents