OpenAI TTS API pricing: tts-1, tts-1-hd and gpt-4o-mini-tts
OpenAI’s speech endpoint turns text into audio with three models, each priced in its own way. This guide lists their prices from OpenAI’s pricing page as of November 2026, what an hour of audio costs, and what changes when these models are switched off in January 2027.
Prices of the three models
tts-1 costs $15 per million characters of input and tts-1-hd $30. gpt-4o-mini-tts is billed by tokens: $0.60 per million text tokens in and $12 per million audio tokens out. OpenAI’s pricing page does not say how many audio tokens make a minute, so the cost of an hour with gpt-4o-mini-tts depends on your audio and is best measured on a real sample.
What an hour costs
At about 1,000 characters a minute, the rule ElevenLabs uses on its pricing page, an hour of speech is about 60,000 characters: about $0.90 with tts-1 and $1.80 with tts-1-hd. The 216 hours that the $5 SuperClonica Starter plan voices at most in a month would cost about $194 with tts-1 and $389 with tts-1-hd.
The January 2027 shutdown
OpenAI’s deprecations page lists tts-1, tts-1-hd and gpt-4o-mini-tts as deprecated: they will be removed from the API on January 6, 2027, and the suggested replacement is gpt-realtime-2.1-mini, priced at $0.60 per million text tokens in and $20 per million audio tokens out. An app built on the speech endpoint needs changes either way, which makes it a good moment to compare prices.
The same hours on SuperClonica
The SuperClonica speech API bills by time, not characters: a plan gives minutes of audio in every 10-minute window. Used without a break, the $5 Starter plan voices up to 216 hours a month, the $10 plan up to 576, $20 up to 1,440 and $50 up to 4,680. A free key gives 3 minutes a day. Every catalog voice and your own trained voices work on any plan, and audio streams back from 300 ms in our measurements. Supertonic 3 speaks 31 languages; Chinese is not among them yet.
When OpenAI is the better choice
OpenAI covers more languages, including Chinese, and its realtime models suit voice agents that also listen and answer in one session. If your app says only a few sentences a day, paying per character may cost less than any monthly plan. SuperClonica pays off when you voice many hours a month, want character or custom voices from the catalog, or need a fixed bill.
Try it in one request
Get a free key on the site and post text to https://superclonica.com/api/v1/speech with the header Authorization: Bearer and your key, and the body
{"text": "...", "voice": "F1", "language": "en"}Add "format": "mp3" for MP3, or "stream": true to play the audio as it arrives.