OpenAI TTS alternatives for real-time voice agents (before the January 2027 shutdown)
Blog post from AssemblyAI
OpenAI announced that tts-1, tts-1-hd, and two gpt-4o-mini-tts snapshots will be retired on January 6, 2027, and recommends gpt-realtime-2.1-mini, which uses the Realtime API rather than the existing /v1/audio/speech endpoint, making the move an architectural change rather than a model-name swap. Teams can instead retain a separate text-to-speech stage by switching providers, with choices including AssemblyAI, ElevenLabs, Cartesia, Inworld, Rime, or a self-hosted model; their documented differences include OpenAI endpoint compatibility, streaming text, interruption handling, telephony audio formats, language coverage, and concurrency limits. The text highlights these features, along with latency, as key considerations for voice agents and notes that provider changes may also affect model and voice names, supported audio formats, speech controls, word timings, and connection management. It suggests testing candidate services with representative scripts and load before the shutdown, while matching the migration path to priorities such as minimizing code changes, supporting live phone calls, staying within OpenAI’s ecosystem, or keeping audio on private infrastructure.
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.