# GPT Realtime Translate

> Native audio-to-audio translation with voice cloning. Preserves original speaker tone. 13 target languages (es/pt/fr/ja/ru/zh/de/ko/hi/id/vi/it/en).

Human version: https://kymaapi.com/models/gpt-realtime-translate
Live JSON: `GET https://kymaapi.com/v1/models` (no auth required)

## Facts

- **Model ID**: `gpt-realtime-translate` — pass this as `model` in the request body
- **Creator**: OpenAI
- **Released**: 2026-05-07
- **Price**: $0.0459 / minute

## Call it

```bash
curl https://kymaapi.com/v1/audio/speech \
  -H "Authorization: Bearer $KYMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-realtime-translate", "prompt": "..."}'
```

## Positioning

GPT Realtime Translate is an audio-to-audio translation model built for low-latency voice conversations. Reach for it when you need to translate spoken input into localized audio while preserving the original speaker's vocal tone.

## About

The model accepts raw audio input and returns translated audio output without intermediate text conversion. It supports voice cloning across 13 languages: Spanish, Portuguese, French, Japanese, Russian, Chinese, German, Korean, Hindi, Indonesian, Vietnamese, Italian, and English.

On Kyma, it operates as a realtime websocket connection billed per minute. Requests automatically failover if a serving path degrades, and every response includes the exact cost in usage.cost alongside an X-Kyma-Model header. Prompt caching bills repeated prompt prefixes at 10% of the input rate.

It is strictly an audio modality model with a 3600-token context window. It does not support text inputs, vision, structured outputs, or explicit reasoning steps, and runs on a premium cost tier with fast inference speeds.

## Use cases

- **Live Cross-Language Calls** — Translates spoken audio in real time while maintaining the original speaker's voice characteristics.
- **Automated Media Dubbing** — Generates localized audio tracks that match the source speaker's tone and pacing.
- **Multilingual Voice Agents** — Powers conversational bots that listen and reply in translated audio without text intermediaries.
- **Real-Time Meeting Translation** — Streams live audio translation for international team discussions and broadcasts.

## Not ideal for

Do not use this model for text-based tasks, vision analysis, structured data extraction, or workflows requiring explicit reasoning steps.

## Pick something else when

- You need text input alongside audio processing → `gemini-3.1-flash-live-preview`
- You require strict JSON or structured output formats → `gemini-2.5-flash-native-audio-preview-12-2025`
- You need a larger context window for long sessions → `gemini-3.5-live-translate-preview`

## See also

- All models: https://kymaapi.com/models.md
- Pricing: https://kymaapi.com/pricing.md
- Other models by OpenAI: https://kymaapi.com/models?q=OpenAI