release
OpenAI GPT Transcribe becomes the default speech-to-text model in Lovable
docs.lovable.devagent
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.
Stacks this affects
Lovable ships in 1 tested stack here.
More Lovable
featureWorkspace extensibility with Events and Hooks APIfeatureEvents & hooks extend Lovable workspaces with custom servicesfeatureLovable workspaces now support Events & Hooks for extending with custom servicesshutdownGemini 2.5 chat models are retired for your app's AI featuresshutdownGemini 2.5 models deprecated and replaced with Gemini 3.8/3.1shutdownGemini 2.5 models retire in favor of Gemini 3.8 and 3.1