Yeah, that difference is real. Retell is faster mainly because they control and optimize the entire pipeline end-to-end, so there’s less delay between speech-to-text, the model response, and text-to-speech. With Vapi, you’re usually chaining multiple providers, which adds extra round trips and latency unless everything is heavily optimized. Also, Retell is very aggressive with streaming and starts responding earlier, which makes it feel much faster even if the model speed is similar. If you want, I can help you tune your Vapi setup to get much closer to that speed @cem802