OpenAI is shutting down tts-1: what breaks and how to move
Apps that send model tts-1 to OpenAI stop speaking on 6 January 2027. If your app has an OpenAI-compatible endpoint setting, you can move by changing the base URL and the key.
OpenAI is shutting down tts-1: what breaks and how to move
Facts checked 16 November 2026. Rival figures come from each rival's own page, linked in the table.
What is being retired, and when
OpenAI's deprecations page lists tts-1, tts-1-hd and
gpt-4o-mini-tts (with its dated snapshots) as deprecated, with shutdown on 6 January 2027 (notice date 1 October
2026). The replacement it names is gpt-realtime-2.1-mini, a Realtime API model, not another model for
POST /v1/audio/speech.
What breaks
Anything that sends model: "tts-1" to OpenAI's /v1/audio/speech after that date: Home Assistant speech
integrations, Obsidian and browser read-aloud plugins, Open WebUI, scripts that make MP3 files. Apps with an
"OpenAI-compatible endpoint" setting keep working if you give them another base URL that speaks the same request.
Moving to Glyphcall
Change the base URL and the key. The request stays the same.
from openai import OpenAI
client = OpenAI(base_url="https://glyphcall.com/v1", api_key="adt_...") # key: https://glyphcall.com/credits (from $1)
with client.audio.speech.with_streaming_response.create(model="tts-1", voice="nova", input="Hello from the new endpoint.") as r:
r.stream_to_file("hello.mp3")
curl https://glyphcall.com/v1/audio/speech -H "Authorization: Bearer adt_..." -H "Content-Type: application/json" \
-d '{"model":"tts-1","voice":"nova","input":"Hello from the new endpoint."}' --output hello.mp3
Home Assistant-style settings: API URL https://glyphcall.com/v1 (or the full https://glyphcall.com/v1/audio/speech
where the app asks for the endpoint), API key adt_..., model tts-1, voice from GET https://glyphcall.com/v1/audio/voices
(free, no key; it fills the voice picker of apps that read it).
What you get, exactly:
- Model: Deepgram Aura-2 on Cloudflare Workers AI.
tts-1is accepted as a name only; this is not OpenAI's model and these are not OpenAI's voices. OpenAI voice names (alloy, ash, coral, echo, fable, nova, onyx, sage, shimmer) are mapped to Aura-2 voices; theX-Voiceresponse header names the voice that spoke. - Limits: up to 2,000 characters per request, one MP3 file back. English (40 voices) by default, Spanish
(10 voices) with
"language": "es". - Price with a prepaid key: 4 credits ($0.004) per started 100 characters, never more than $0.01 up to 300
characters or $0.08 up to 2,000. The
X-Credits-Chargedheader states it. Failed calls are not charged. - Price per call without a key (x402 or MPP, USDC): $0.01 up to 300 characters, $0.08 up to 2,000.
- Free test key:
POST https://glyphcall.com/keys/voice-testgives 25 calls (speech up to 100 characters, audio up to 30 seconds) when the offer is on; if it answers 404 the offer is off and a $1 key is the way in. - Errors come in OpenAI's shape,
{"error": {"message", "type", "code"}}, and the message is a sentence.
What we do not support: response_format other than mp3 (no opus, aac, flac, wav, pcm), speed other than 1,
instructions, a separate HD tier (tts-1-hd is accepted and gives the same voices), streaming of partial audio,
realtime speech, voice cloning, and languages other than English and Spanish on this route.
Test the key first
If something does not work in the app, check the key and the endpoint with one call from a terminal:
curl https://glyphcall.com/v1/audio/speech \
-H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" \
-d '{"model":"tts-1","input":"Hello from Glyphcall Voice.","voice":"nova"}' -o hello.mp3
What it costs
| Call | Price |
|---|---|
| Text to speech, up to 300 characters | $0.01 (10 credits) |
| Text to speech, 301 to 2,000 characters | $0.08 (80 credits) |
| Speech to text, a file up to 5 MB | $0.01 (10 credits) |
| Speech to text, a larger MP3 or WAV up to 60 minutes / 25 MB | $0.05 (50 credits) |
A failed call is never charged. No account, no subscription unless you choose the monthly plan.
Questions
When does tts-1 stop working? OpenAI's deprecations page gives 6 January 2027 as the shutdown date for tts-1, tts-1-hd and gpt-4o-mini-tts.
Do I have to change my code? Only the base URL and the key. The request body (model, voice, input) stays the same; MP3 is the only output format.
Other guides: Open WebUI · LibreChat · AnythingLLM · Home Assistant · Obsidian Whisper · Any other app · OpenAI whisper-1 shutdown · Voice list