AI agent seems to stop work after a minute - rate limiting?

[yannick_jojo]

Hello

I'm having a fun time experimenting with the AI stuff, but it seems that the bot stops anwsering after a minute.

Is this some kind of rate limiting?

How do I turn it off?

 

[yannick_jojo]

I did a RTFM and it clearly states " Under Rate limits, increase limits for the models above as needed (larger installations may require higher limits to avoid unpredictable behavior).". So excuse me . What would be a good limit?

 

[yannick_jojo]

These are the current limits, but it seems I can not override them. I can only limit them.

 

[KyriacosS_3CX]

Hi,

Can you check PBX and OpenAI logs to see what info you are getting on this, whether you are hitting a limit or it's something else?

 

[yannick_jojo]

Where in 3CX would I find logs regarding bots?

 

[Kevin Attard Compagno]

The 3CX Support Info file will contain a "3CXAI.log" file. But note also as @KyriacosS_3CX mentioned - look at OpenAI logs to check what is happening on the OpenAI side of things.

 

[yannick_jojo]

The OpenAI logs seem more logical to troubleshoot, but it looks pretty empty. From what I can find I need to look here: https://platform.openai.com/settings/organization/usage

But I don't really see any information besides the cost:

 

[nikolascx]

Hallo @yannick_jojo
What realtime model are you using?
Also you are in the wrong page. there are 2 limits in the tree - one is for Organization limits and the other is for the Project limits.
You need to go in the Project Limits (bottom)



then check this url
https://www.3cx.com/docs/ai-agent-configuration/

Just leave all models, but before you touch limits, make sure you are reaching your limits.
What is your current value?

One minute of continuous audio conversation consumes approx 18k-20k tokens.
If you have 40k and you are the only one testing, and making one call at a time, you are well in the limit.

Make a call, and reproduce the problem and send the logs in Private and I will check for you.

 

[KyriacosS_3CX]

Hi All,

We just published a Blog post about avoiding rate limiting for the AI Agents. You can read it here.