The current API architecture provided by OpenAI cannot be applied in certain scenarios:

• Network propagation delay will completely ruin applications such as autonomous driving.
• For some banking-related operations, do I really need to send passwords and other private data to OpenAI's API servers?
• Additionally, there are scenarios involving edge computing, local computing, and offline environments.