Realtime API "rate_limit_exceeded" "We're currently processing too many requests — please try again later."

API
1

When I hit the Realtime API using the Playground and the Realtime Console I get a conversation.item.input_audio_transcription.failed message. Looking into the response.done object I see the following:

type:"response.done"
event_id:"event_AE(...)"
response.object:"realtime.response"
response.id:"resp_AE(...)"
response.status:"failed"
response.status_details.type:"failed"
response.status_details.error.type:"invalid_request_error"
response.status_details.error.code:"rate_limit_exceeded"
response.status_details.error.message:"We're currently processing too many requests — please try again later."
response.usage.total_tokens:125
response.usage.input_tokens:125
response.usage.output_tokens:0
response.usage.input_token_details.cached_tokens:0
response.usage.input_token_details.text_tokens:118
response.usage.input_token_details.audio_tokens:7
response.usage.output_token_details.text_tokens:0
response.usage.output_token_details.audio_tokens:0

My rate limits are clear. I have credits. I can’t tell if this is an issue with my account or if it’s a system issue.

2

Although it’s a bit ambiguous I believe:

The error message isn’t saying that you are hitting the rate limit.

Instead, it’s saying that the server is over capacity right now.

3

That’s what I figured. https://status.openai.com/ says everything is nominal though. How do I raise a flag so they look into what’s happening?

4

I’d imagine they’re aware and why they’re very slowly rolling out access.

Will probably be a bumpy ride for the next couple days.

1 Like
5

image
image919×900 61 KB

6

Yeah. Now I’m getting rate limit messages on the entire API:

"You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors."

This looks a lot like this incident from yesterday: OpenAI Status - Erroneous usage data for some API users is leading to them incorrectly hitting usage limits

Hopefully someone is on it.

7

I found a workaround of sorts: Even though I’m on auto-pay, the system was showing me at negative $0.04. I manually bumped it up to $50 and now it works again. Never had to do that before, and it worked fine on Tuesday, so… weird.

8

That is weird. Thanks for reporting the solution though.

9

So, they have a tiered system. If your account has 5$ - tier one, 50$ is tier 2 and so on see.

Now, the tiers, tell you how many requests you can send per minute. To see your tier and how many request you can send per day or per minute see limits.
Hope it helps. For now take note, voice is still pretty expensive because each request can cost you some cents. Try to keep testing minimal by looking at various code examples for your language of choice.