ReferenceAsk
Back to Remy
Remy Reference/models/GPT Realtime 2.1 Mini TTS
O

GPT Realtime 2.1 Mini TTS

GPT Realtime 2.1 Mini TTS, a text-to-speech model from OpenAI.
speechModelOverride: { model: "gpt-realtime-2.1-mini-tts" }
OpenAI·Released July 2026
Pricing
Text input
$0.60 / 1M
Cached input
$0.060 / 1M
Text output
$2.40 / 1M
Audio output
$20.00 / 1M
Specifications
Context window
1,500 tokens
Released
July 2026
Model string
gpt-realtime-2.1-mini
Parameters
Model options, passed in speechModelOverride.config.
ParameterTypeDefaultRange / options
voiceSelectmarinmarincedaralloyashballadcoral+4
instructions
Optional direction for the delivery: tone, pace, accent or character (e.g. "calm and reassuring, like a late-night radio host").
Prompt——
speed
Speaking speed. 1.0 is normal.
Number10.25–1.5, step 0.05
Voices · 10
Marin
Cedar
Alloy
Ash
Ballad
Coral
Echo
Sage
Shimmer
Verse
Use this model
await mindstudio.textToSpeech({
  text: "...",
  speechModelOverride: {
    model: "gpt-realtime-2.1-mini-tts",
    config: {
      voice: "marin",
      speed: 1,
    },
  },
});