エンタープライズ機能を備えた OpenAI の GPT-3 モデルへのアクセスを提供する Azure サービス。
Hi 津﨑 友一郎
Could you share your observation in East US and Japan East. As per my trials, Japan East looked bit slower but looks Ok to operate (27 second in Japan East 3334 token, 20 second 2755 token in East US)
Considering the fact, we are using GPT 5 reasoning models, with streaming response, response time look reasonable to me as per last experience with product group on this.
Confirmed with product group.
If you have any counter observation with respect to prompt size and TPM configuration. Please share the details requested via private message.
Thank you for using Azure Services.