@Emanuele You might want to enable monitoring on your azure OpenAI resource to check for latency. You can also report the behavior by providing details about your resource like model, region, version, scenario and no. of tokens used. This would help the team to check if there is any service issues that could increase your deployment response times. Thanks!!
If this answers your query, do click Accept Answer
and Yes
for was this answer helpful. And, if you have any further query do let us know.