Overview
GPT-Realtime mini is a cost-efficient version of GPT-Realtime for real-time speech-to-speech applications. OpenAI announced its release on October 6, 2025. (developers.openai.com)
Capabilities and uses
- Accepts text, image and audio input; produces text and audio output. It can respond in real time over WebRTC, WebSocket or SIP, making it suitable for interactive voice assistants and other real-time conversational applications.
- Supports function calling through the Realtime API. The model page lists a 32,000-token context window, a 4,096-token maximum output and an October 1, 2023 knowledge cutoff. (developers.openai.com)
Limitations and availability
- Video, structured outputs, fine-tuning and the Batch API are not supported. (developers.openai.com)
- The release snapshot
gpt-realtime-mini-2025-10-06was shut down on July 23, 2026. The model aliasgpt-realtime-mininow points to the2025-12-15snapshot and is deprecated, with shutdown announced for January 20, 2027. These are different availability statuses. (developers.openai.com)
Pricing
OpenAI API list prices for the gpt-realtime-mini alias, checked November 17, 2026, are in USD per 1 million tokens. They are not verified as the historical prices of the shut-down October snapshot. (developers.openai.com)
| Modality | Input | Cached input | Output |
|---|---|---|---|
| Text | $0.60 | $0.06 | $2.40 |
| Audio | $10.00 | $0.30 | $20.00 |
| Image | $0.80 | $0.08 | Not supported |