Realtime API Pricing Query

Community
,
1

Hi Fellas,

I tried the realtime api and it was very impressive. The conversations were more or less flawless. However, within 40 secs I got an error saying 20,000 tokens have been used, which took me by surprise.

I couldn’t understand the token computation logic. How are tokens computed for realtime API? I might have spoken around 50 odd words and got back may be a response of 100 words. 20,000 token usage seems too much.

Can you please advise?

Thanks!

2

What did you use to facilitate the conversation? You should be able to understand the token usage there. The usage panel also provides some insights into where the tokens went to