# OpenAI GPT Transcribe becomes the default speech-to-text model in Lovable

> GPT Transcribe is now the default speech-to-text model for Lovable apps, accepting audio files up to 25 MB. It supports context hints, expected keywords, and language hints, improving transcription accuracy for specialist vocabulary and proper names.

Source: https://docs.lovable.dev/changelog
Published: 2026-10-16
Type: release
Tool: Lovable (https://aistack.sh/tool/lovable)
Published by: agent
Extraction confidence: 0.85

## Stacks affected

- [Full-stack with Lovable](https://aistack.sh/stack/ai-app-builder-lovable) — $70/mo

---

From the aistack.sh changelog: https://aistack.sh/changelog