I’m very confused that what is main difference between gpt-realtime and gpt -realtime-mini and also in pricing page thier is pricing in token but how many token it will cost per minute can you please provide me rough estimate
gpt-realtime is the more powerful and faster model, so it costs more tokens. gpt-realtime-mini is a lighter version that is cheaper and a bit less capable. The main difference is performance vs cost.
can you guide me that how much token both models will consum per minutes so that I can estimate cost
I think you’ve got it figured out - but that’s how many tokens I get per minute.
Calculate for yourself how much is in regular and how much is in mini.
(By the way, I tried the mini and the regular - and I really don’t recommend the mini - very bad experience)
In my bot - most of the time the bot talks, if your usage is different - the token calculation will be completely different, and I have a fairly long prompt - and it carries it with every answer - it also affects the tokens.
Input tokens:
• Text (new): 23,119
• Text (cached): 112,512
• Total text: 135,631 tokens
• Audio (new): 686
• Audio (cached): 2,048
• Total audio: 2,734 tokens
• Total cached: 114,560 tokens
• Total input: 138,365 tokens
Output tokens:
• Text: 662 tokens
• Audio: 666 tokens
• Total output: 1,328 tokens