Changes
- OpenAI text-to-speech supports Realtime models and new agents default to
gpt-realtime-2.1-mini(voicecedar). OpenAI shuts downtts-1,tts-1-hdandgpt-4o-mini-ttson 2027-01-06: switchtext_to_speech.model_idtogpt-realtime-2.1-miniand pick a Realtime voice (onyx,novaandfableare not available). (e59b082)
Full changes: v2026.11.02.0...v2026.11.02.1