Dear Vapi Team, I hope this message finds you well. I am encountering an issue where my assistant rarely hangs up on its own. I have reviewed the prompt setup and checked the logs, and I noticed the following behavior: [CHECKPOINT] Model sent end token This indicates that the AI seems to understand the intent to end the conversation. However, the disconnection doesn’t occur as expected. I’d like to know if there’s a configuration I can adjust on my side, or if this might be an issue on your end. For context, I am using Vapi both via the web snippet on our page and through the dashboard’s "Talk to Assistant" feature. Your product is fantastic, and we see significant potential in its capabilities. I’d appreciate your guidance in resolving this issue. Best regards, Patrick

message has been deleted

Thank you for your quick answer, do you know when it comes out, it looks like it could come soon, but I'm not sure

Hey @DerUnglaublicheToast To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help me figure out what went wrong!

@Shubham Bajaj Sorry for the last answer that I deleted I had a other request and I mixed it in my head. d0bdfbd3-d825-4bf9-a319-e3426b9c58c4 but it's on every bot that I created I expected that the assistant ends the call by itself But instead he says basically good bye and nothing happend in die Log he even sends an endtoken but as you see the call continues Log Example Snippet: 11:01:34:759 [LOG] Voice input: Ich werde jetzt das Gespräch beenden. 11:01:34:765 [LOG] Model output: önen 11:01:34:771 [LOG] Model output: Tag 11:01:34:778 [LOG] Model output: noch 11:01:34:783 [LOG] Model output: ! 11:01:34:797 [LOG] Model request cost (attempt #1, $0.0003756, 2260 prompt, 61 completion) 11:01:34:797 [LOG] Voice input: Schönen Tag noch! 11:01:34:797 [CHECKPOINT] Model sent end token 11:01:34:872 [CHECKPOINT] Voice audio received 11:01:34:881 [INFO] Turn latency: 1333ms (transcriber: 199ms, endpointing: 102ms, model: 423ms, voice: 600ms) 11:01:34:881 [CHECKPOINT] Assistant speech started 11:01:48:204 [CHECKPOINT] Assistant speech stopped 11:01:53:647 [CHECKPOINT] User speech possibly starting 11:01:53:786 [CHECKPOINT] User speech started 11:01:53:866 [CHECKPOINT] User speech possibly stopping 11:01:54:078 [LOG] Endpointing timeout 100ms (rule:

heuristic

) 11:01:54:078 [LOG] Transcriber output: Tschüss. 11:01:54:179 [CHECKPOINT] Model request started 11:01:54:243 [LOG] Model request started (attempt #1, gpt-4o-mini-2024-07-18, azure-openai, southcentralus) 11:01:54:406 [CHECKPOINT] User speech stopped 11:01:54:614 [CHECKPOINT] Model sent start token 11:01:54:616 [CHECKPOINT] Model sent first output token 11:01:54:616 [LOG] Model output: war 11:01:54:616 [LOG] Model output: Es 11:01:54:629 [LOG] Model output: schön 11:01:54:629 [LOG] Model output: , 11:01:54:632 [LOG] Model output: mit 11:01:54:644 [LOG] Model output: dir 11:01:54:652 [LOG] Model output: zu 11:01:54:699 [LOG] Model output: sprechen

Curently I away from keyboard will reply back to you in few hours.

@Shubham Bajaj Ok thank you

@DerUnglaublicheToast you have shared the assistant id not the call id. If you can share the call id then it will be helpful for me.

@User Sorry I miss understood, here are my call ids 1d2d55ae-aa01-40c3-89aa-eee160a2b97c I think this one displays the issue better: 816786fe-70b4-4da9-8e28-e8c025b42bf3

@DerUnglaublicheToast did you tried without KB? BTW the new KB developments has been pushed try it out as well, and let me know how it goes!!

Knowledge base ? I will try ist without it and let you know afterwards

@DerUnglaublicheToast do let me know when your try out, and can I close this ticket for now?