Hi all,
I have seen a few threads on the forums about understanding the cost of the Realtime API, but no concrete answers, so hoping for some clarification. Here I am posting my usage + cost for a ~5 minute conversation.
Here’s my expected pricing calculation based on https://openai.com/api/pricing/
Audio input → $100/1m tokens
Audio output → $200/1m tokens
Cached → $20/1m tokens
Input (3243 tokens): $0.48
Output (2173 tokens): $0.43
Cache (4864 tokens): $0.1
Total: $0.92
However, the actual charge in 3.5x higher. I am seeing this discrepancy continuously between the usage and cost across different accounts as well. I understand that lots of tokens can be consumed when interrupting, etc, but presumably those would show up in the usage. Thoughts / clarifications would be greatly appreciated.