Im getting about 500-800 ms on the attributes.mode...
# general-english
k
Im getting about 500-800 ms on the attributes.model.startTokenLatencyMs on gpt 5.1.