OpenAI deprecated GPT-5.3-Codex, GPT-5.1 and GPT-5.4-Nano, along with its older text-to-speech models, on October 1, 2026, according to its API deprecations page. The language models leave the API on April 1, 2027, and the speech models on January 6, 2027.
On this page
Language models: shutdown on April 1, 2027
OpenAI says the three models “are deprecated and will be removed from the API on April 1, 2027, with six months’ notice.” Each has a named replacement in the GPT-6 family: gpt-6-sol for the two larger models and gpt-6-luna for the nano model.
Text-to-speech models: shutdown on January 6, 2027
A second notice dated the same day covers speech generation. OpenAI says these models “are deprecated and will be removed from the API on January 6, 2027, with at least three months’ notice,” and directs developers to migrate to gpt-realtime-2.1-mini. The notice covers tts-1, tts-1-hd and two dated snapshots of gpt-4o-mini-tts.
Full list
| Model | Shutdown date | Replacement |
|---|---|---|
gpt-5.3-codex | April 1, 2027 | gpt-6-sol |
gpt-5.1 | April 1, 2027 | gpt-6-sol |
gpt-5.4-nano | April 1, 2027 | gpt-6-luna |
tts-1 | January 6, 2027 | gpt-realtime-2.1-mini |
tts-1-hd | January 6, 2027 | gpt-realtime-2.1-mini |
gpt-4o-mini-tts-2025-03-20 | January 6, 2027 | gpt-realtime-2.1-mini |
gpt-4o-mini-tts-2025-12-15 | January 6, 2027 | gpt-realtime-2.1-mini |
whisper-1 | February 26, 2027 | gpt-live-transcribe or gpt-transcribe |
gpt-4o-transcribe | February 26, 2027 | gpt-live-transcribe or gpt-transcribe |
gpt-4o-mini-transcribe | February 26, 2027 | gpt-live-transcribe or gpt-transcribe |
gpt-4o-transcribe-diarize | February 26, 2027 | gpt-live-transcribe or gpt-transcribe |
The last four rows come from an earlier notice, described below.
Context: transcription models were deprecated in August
The October notices follow an August 26, 2026 deprecation of OpenAI’s speech-to-text line. The OpenAI API changelog records that OpenAI “announced the deprecation of whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, and gpt-4o-transcribe-diarize,” with shutdown on February 26, 2027. The deprecations page lists gpt-live-transcribe or gpt-transcribe as the replacement for each.
Taken together, the two rounds move both directions of OpenAI’s audio API to newer models in notices issued about five weeks apart: transcription to the gpt-transcribe models by late February 2027, and speech output to gpt-realtime-2.1-mini by early January 2027.
What it changes
Applications that call any of the listed model names keep working until the shutdown date, after which OpenAI removes the models from the API. The text-to-speech deadline comes first: developers have about three months from the notice to move speech output to gpt-realtime-2.1-mini, compared with six months for the three language models.
Current and retired OpenAI models are tracked in our AI model release timeline, and the live state of OpenAI’s platform is on the OpenAI API status page.
