OpenAI will remove tts-1, tts-1-hd and gpt-4o-mini-tts on January 6, 2027

On October 1, 2026, OpenAI deprecated the four text-to-speech models in its API. They will be removed on January 6, 2027. OpenAI’s deprecations page tells developers to move to gpt-realtime-2.1-mini before that date and links to its Realtime API guide.

Which models are going

Model	Removed on	Recommended replacement
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

What it costs now

OpenAI’s pricing page, as we read it on October 18, 2026:

Model	What is billed	Price per 1M
tts-1	Characters of text	$15.00
tts-1-hd	Characters of text	$30.00
gpt-4o-mini-tts	Text input tokens	$0.60
gpt-4o-mini-tts	Audio output tokens	$12.00
gpt-realtime-2.1-mini	Text input tokens	$0.60
gpt-realtime-2.1-mini	Audio output tokens	$20.00

The tts-1 models bill by character and the replacement bills by token, so the pages don’t give a like-for-like cost per minute of speech. The only way to know is to run your own text through both. gpt-realtime-2.1-mini also lists audio input at $10.00 per 1M tokens, for apps that send speech in.

Part of a wider audio clean-up

This is the latest of several audio notices on the same page. On July 20, 2026, OpenAI deprecated older realtime and audio models such as gpt-audio and gpt-realtime, with removal on January 20, 2027. On August 26, 2026, it deprecated whisper-1 and the gpt-4o-transcribe family, with removal on February 26, 2027.

What to do

Search your code for the four model IDs above. Test gpt-realtime-2.1-mini on your own scripts and voices before December, so you can compare sound and cost while the old models still work.