AI Gateway now supports streaming transcription

Short summary

Vercel's AI Gateway now supports streaming transcription in beta, enabling real-time audio-to-text with low latency for live captioning and voice input. The feature works through the AI SDK's streamTranscribe function and supports multiple providers including OpenAI's GPT-Realtime-Whisper and xAI's Grok-STT. It also enables easy voice mode integration for AI agents by streaming user speech to a transcription model and passing live text to any text-based agent.

• •AI Gateway adds beta streaming transcription via AI SDK's streamTranscribe
• •Supports cross-provider models including OpenAI Whisper and xAI Grok-STT
• •Enables low-latency live captioning, voice input, and voice mode for AI agents

Generated with AI, which can make mistakes.