That error means the Vapi execution worker handling the call is dying mid-session. The cause might be, Memory exhaustion (container OOM kill). Infinite loop / unhandled promise in pipeline. Voice provider timeout crashing the worker. Bad deploy (corrupt build or missing env vars). Rate limit spike causing worker instability. Since it’s 100%, this is likely, A bad deployment running on all workers. A config/env variable issue affecting startup. Platform-side worker outage You can check, Verifying current deployed version hash. Check container logs for OOM / crash stack traces. Confirm all required env vars are present. Restart workers cleanly (not rolling restart full stop/start). If self-hosted, check memory + CPU usage. I can guide you step-by-step to isolate whether this is infra, deployment, or pipeline logic and get calls stable again fast.