I’ve noticed the distortion issue with the GPT real-time model when using a free Vapi number too. I’m tracking it and can help troubleshoot or implement workarounds. If there’s an official fix planned, I’ll share updates as soon as they’re available. @So[U]L