correct-apricot•3d ago

Unexpected Multi-LLM Usage Causing High Call Costs

Hi Vapi team, We’ve noticed an issue where one of our assistant calls was unexpectedly routed through multiple LLM providers, which caused unusually high costs. Details: Call ID: 653fbd91-fa0b-4758-a6dd-b067947b6f8f Org ID: 21c685b6-4054-43c6-8a20-157334ca216e Date/Time: Aug 26, 01:30 UTC Duration: ~2 minutes Cost: $2.39 (unexpectedly high for such a short call) Issue Observed: From the logs, I can see that during this single call, Vapi made requests to different providers/models: azure-openai → gpt-4-1106-preview azure-openai → gpt-4-0125-preview openai → chatgpt-4o-latest This looks like multiple model invocations happened for the same call, instead of sticking to a single provider/model configuration. Could you please investigate why this call routed through multiple LLMs?