OpenAI speech, audio, and realtime shutdowns in January 2027
The dataset has 5 OpenAI ids with shutdown date 2027-01-06: gpt-4o-mini-tts, gpt-4o-mini-tts-2025-03-20, gpt-4o-mini-tts-2025-12-15, tts-1, tts-1-hd. The deprecations page says to migrate those speech models to gpt-realtime-2.1-mini. The text-to-speech page says that id is not a supported model for /v1/audio/speech. The dataset has 13 OpenAI ids with shutdown date 2027-01-20: gpt-4o-audio, gpt-4o-mini-audio, gpt-4o-mini-realtime, gpt-4o-mini-transcribe-2025-03-20, gpt-4o-realtime, gpt-audio, gpt-audio-2025-08-28, gpt-audio-mini, gpt-audio-mini-2025-12-15, gpt-realtime, gpt-realtime-2025-08-28, gpt-realtime-mini, gpt-realtime-mini-2025-12-15. OpenAI's deprecations page says: "At the time of the shut down, the model or endpoint will no longer be accessible."
Last reviewed . Publisher: SunsetDiff.
Steps
Split the two dates. Speech ids in the first table shut down on January 6, 2027. Audio, realtime, and gpt-4o-mini-transcribe-2025-03-20 shut down on January 20, 2027. A snapshot with a different stored date is not in these tables.
Find the ids in your code. Search for the short id and for each dated snapshot. A speech call, a chat audio call, and a realtime session can live in different files.
Change the call, not only the string. For tts-1, tts-1-hd, and gpt-4o-mini-tts, the text-to-speech page says to move to gpt-realtime-2.1-mini on the Realtime API. gpt-audio-1.5 stays on Chat Completions. gpt-realtime-2.1 stays on the Realtime API. Do this before the date on that row.
Speech models shutting down on January 6, 2027
These rows are the OpenAI ids in the dataset with shutdown date 2027-01-06. The model id links to its SunsetDiff model page. Deprecation date, shutdown date, and replacement are the fields stored on that row.
OpenAI's deprecations page says: "The following text-to-speech models are deprecated and will be removed from the API on January 6, 2027, with at least three months' notice. Migrate to gpt-realtime-2.1-mini before the shutdown date. See the Realtime API guide to plan your migration." The table lists tts-1, tts-1-hd, and gpt-4o-mini-tts with snapshots gpt-4o-mini-tts-2025-03-20 and gpt-4o-mini-tts-2025-12-15. Every replacement cell on that table is gpt-realtime-2.1-mini.
The same deprecations page uses January 6, 2027 for a different change: "Active existing customers will no longer be able to create new fine-tuning jobs on this date. Inference on fine-tuned models will be disabled only when the underlying base model is deprecated." That sentence is about new fine-tuning jobs. It is not an extra model id, so it is not a row in this table.
5 OpenAI speech ids with shutdown date 2027-01-06.
Audio and realtime models shutting down on January 20, 2027
These rows are the OpenAI ids in the dataset with shutdown date 2027-01-20.
OpenAI's deprecations page says: "On July 20, 2026, we notified developers using legacy audio, realtime, and transcription model families and snapshots of their deprecation and removal from the API on January 20, 2027." That table lists gpt-realtime with snapshot gpt-realtime-2025-08-28, gpt-audio with snapshot gpt-audio-2025-08-28, gpt-4o-audio, gpt-4o-realtime, gpt-realtime-mini with snapshot gpt-realtime-mini-2025-12-15, gpt-audio-mini with snapshot gpt-audio-mini-2025-12-15, gpt-4o-mini-realtime, gpt-4o-mini-audio, and gpt-4o-mini-transcribe-2025-03-20.
gpt-realtime-mini-2025-10-06 and gpt-audio-mini-2025-10-06 are not in this table. Their stored shutdown date is 2026-07-23. gpt-4o-mini-transcribe and gpt-4o-mini-transcribe-2025-12-15 are not in this table. Their stored shutdown date is 2027-02-26. Those ids, plus whisper-1, gpt-4o-transcribe, and gpt-4o-transcribe-diarize, are in the February 26, 2027 transcription guide.
13 OpenAI audio, realtime, and transcription ids with shutdown date 2027-01-20.
On January 6, 2027 the speech ids reach the shutdown date stored in the dataset. On January 20, 2027 the audio, realtime, and transcription ids in the second table do. OpenAI's deprecations page says: "At the time of the shut down, the model or endpoint will no longer be accessible." The same page uses sunset and shut down for that moment. It does not print an HTTP status for it.
A 400 from a field the replacement rejects is a different failure. This guide does not invent an error body. The text-to-speech page says gpt-realtime-2.1-mini is not a supported model for /v1/audio/speech. It does not print the error sentence for that call.
Search the repository for every id in both tables. A speech helper, a chat completions audio call, and a realtime session often use different files. How to find every model ID in your codebase covers SDK argument names. The scan page builds a search from dataset ids that have a shutdown date.
OpenAI's deprecations page says to call the List models endpoint, GET /v1/models, and read each model's optional shutdown_date field. When present, the field is a YYYY-MM-DD string, or null if no shutdown date has been announced. The same field is on the Retrieve a model endpoint.
You can also read one id from the free SunsetDiff lookup file. tts-1 is a January 6 row. gpt-4o-realtime is a January 20 row. The API page lists the other files. No key and no signup. The alerts page is the free email form. That page says what the form stores and the one welcome email it can send.
The text-to-speech page says: "The tts-1, tts-1-hd, and gpt-4o-mini-tts models are deprecated and will shut down on January 6, 2027. Migrate to gpt-realtime-2.1-mini using the Realtime API. This is a change of API workflow: gpt-realtime-2.1-mini isn't a supported model for /v1/audio/speech." Existing integrations can keep using the Speech API samples on that page until the shutdown. The replacement is not a new model string on POST /v1/audio/speech.
The gpt-realtime-2.1-mini model page lists v1/realtime as supported. It lists v1/chat/completions, v1/audio/speech, and v1/audio/transcriptions as not supported. The Realtime getting started page shows a RealtimeSession whose model field is gpt-realtime-2.1. For these speech rows, the deprecations page and the text-to-speech page name gpt-realtime-2.1-mini. Use the id from the table.
The text-to-speech page says: "The tts-1 and tts-1-hd models support a smaller set: alloy, ash, coral, echo, fable, onyx, nova, sage, and shimmer." The realtime conversations page says: "Current voice options are alloy, ash, ballad, coral, echo, sage, shimmer, verse, marin, and cedar." That realtime list does not include fable, nova, or onyx. Those three names are in the speech-endpoint list and in the smaller tts-1 and tts-1-hd set. The realtime conversations page also says: "Once the model has emitted audio in a session, the voice cannot be modified for that session."
The deprecations page names gpt-audio-1.5 for gpt-4o-audio, gpt-4o-mini-audio, gpt-audio, gpt-audio-2025-08-28, gpt-audio-mini, and gpt-audio-mini-2025-12-15. The gpt-audio-1.5 model page says Chat Completions at v1/chat/completions is supported. It lists v1/realtime and v1/audio/speech as not supported. Keep the chat completions call and change the model id. Do not send gpt-audio-1.5 to the Realtime API.
The deprecations page names gpt-realtime-2.1 for gpt-4o-realtime, gpt-realtime, and gpt-realtime-2025-08-28. It names gpt-realtime-2.1-mini for gpt-4o-mini-realtime, gpt-realtime-mini, and gpt-realtime-mini-2025-12-15. The gpt-realtime-2.1 model page lists v1/realtime as supported, and lists v1/chat/completions and v1/audio/speech as not supported. It says GPT-Realtime-2.1 supports speech-to-speech interactions with configurable reasoning effort. That page does not print a parameter that gpt-realtime must drop.
The deprecations page names gpt-4o-mini-transcribe-2025-12-15 as the replacement for gpt-4o-mini-transcribe-2025-03-20. That replacement snapshot, and the alias gpt-4o-mini-transcribe, shut down on 2027-02-26. The February 26, 2027 transcription guide covers those ids. The file transcription page says: "whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, gpt-4o-transcribe-diarize, and /v1/audio/translations are deprecated and will shut down on February 26, 2027." For the models that page recommends after that later date, it says: "For gpt-transcribe, languages replaces the singular language field. Don't send both fields." That languages field is documented for gpt-transcribe. The page does not say the December 15 snapshot rejects language.
OpenAI's deprecations page says the notice period is time to evaluate the recommended replacement, test application behavior, and finish the migration before the model is no longer available. For a speech id, test gpt-realtime-2.1-mini on the Realtime API before January 6, 2027. For an audio or realtime id, test the replacement in the second table before January 20, 2027.
The pre-migration test checklist is the longer list: evals, rejected fields, token limits, tools, price, and latency.
The dataset lists gpt-4o-mini-tts, gpt-4o-mini-tts-2025-03-20, gpt-4o-mini-tts-2025-12-15, tts-1, tts-1-hd with shutdown date 2027-01-06. The deprecations page says: "The following text-to-speech models are deprecated and will be removed from the API on January 6, 2027, with at least three months' notice. Migrate to gpt-realtime-2.1-mini before the shutdown date. See the Realtime API guide to plan your migration."
Why is the speech replacement a realtime model?
The deprecations page names gpt-realtime-2.1-mini for tts-1, tts-1-hd, and gpt-4o-mini-tts, including both dated snapshots. The text-to-speech page says: "The tts-1, tts-1-hd, and gpt-4o-mini-tts models are deprecated and will shut down on January 6, 2027. Migrate to gpt-realtime-2.1-mini using the Realtime API. This is a change of API workflow: gpt-realtime-2.1-mini isn't a supported model for /v1/audio/speech." The gpt-realtime-2.1-mini model page lists v1/realtime as supported and v1/audio/speech as not supported.
Which ids shut down on January 20, 2027?
The dataset lists gpt-4o-audio, gpt-4o-mini-audio, gpt-4o-mini-realtime, gpt-4o-mini-transcribe-2025-03-20, gpt-4o-realtime, gpt-audio, gpt-audio-2025-08-28, gpt-audio-mini, gpt-audio-mini-2025-12-15, gpt-realtime, gpt-realtime-2025-08-28, gpt-realtime-mini, gpt-realtime-mini-2025-12-15 with shutdown date 2027-01-20. gpt-realtime-mini-2025-10-06 and gpt-audio-mini-2025-10-06 use 2026-07-23. gpt-4o-mini-transcribe and gpt-4o-mini-transcribe-2025-12-15 use 2027-02-26.
What does OpenAI say happens at shut down?
OpenAI's deprecations page says: "At the time of the shut down, the model or endpoint will no longer be accessible." That page does not print an HTTP status for that moment.