Each response calls the model once; generation can take a minute or two, especially through a relay.
← History