OpenAI releases gpt-realtime-mini-2025-12-15 speech-to-speech model
Action Required
Developers can now leverage a more accurate and reliable speech-to-speech model for real-time applications, improving the quality of voice-based interactions and enabling new use cases.
AI Impact Summary
OpenAI is releasing a new speech-to-speech model, gpt-realtime-mini-2025-12-15, offering feature parity with the full gpt-realtime model for instruction-following and function-calling. This update focuses on API-only deployment with both audio input and output, and includes significant improvements in accuracy, robustness, and multilingual performance, particularly for Japanese, Indic, and other languages. The model also reduces hallucinations in noisy environments, making it suitable for real-world audio streams.
Affected Systems
- Date
- Date not specified
- Change type
- capability
- Severity
- high