Latency much higher / Response timer slower than using webui directly

[bennofs]

Has anyone else observed that using this API, it takes much longer for the model to answer than if using the chat interface directly? Any ideas why this happens?

I verified that the same servers are used and also used Wireshark to see that it really is the server that answers more slowly, and not some delay in processing responses in the client.

[d-Dice]

Especially image generations