I've noticed gpt-4o failing frequently in my workflow. Responses get generated multiple times and spliced together. Example from call id e8901da2-eebd-4ff5-855d-5c8fefb10f76: AI Thank you, Mark! My name is Dakota, and I’m calling from Sequin Financial. Is this a good time to talk?Hi, Mark! My name is Dakota, and I’m calling from Sequin Financial. Is this a good time to talk?Hi, Mark! My name is DakotFor verification purposes, can you verify your loan number and date of birth, or last payment? Here are some relevant logs from the dashboard: 19:12:15:993 [WARN] Model request failed (attempt #1, reason: providerfault-model-hung) 19:12:15:993 [LOG] Model request started (attempt #3, gpt-4o-2024-08-06-global, azure-openai, westus) 19:12:15:993 [LOG] Model request cost (attempt #1, $0.00027, 0 prompt, 27 completion) 19:12:16:027 [CHECKPOINT] Voice audio received 19:12:16:051 [INFO] Turn latency: 3531ms (transcriber: 243ms, endpointing: 1487ms, kb: n/a, model: 1143ms, voice: 632ms) 19:12:16:052 [CHECKPOINT] Assistant speech started 19:12:17:620 [LOG] Model request started (attempt #4, gpt-4o-2024-11-20-global, azure-openai, eastus) 19:12:17:620 [LOG] Model request cost (attempt #3, $0, 0 prompt, 0 completion) 19:12:17:620 [WARN] Model request failed (attempt #3, reason: providerfault-model-hung) 19:12:17:753 [CHECKPOINT] Assistant speech stopped 19:12:19:574 [WARN] Model request taking too long (attempt #4, started 1953.33ms ago) 19:12:19:574 [LOG] Model request started (attempt #5, gpt-4o-2024-11-20-global, azure-openai, westus)

Thank you for providing the call id. We are reviewing the details of your request and will reach out to you when we have an update or if we need additional information.