Não decore palavra por palavra: entenda a estrutura da resposta e refaça com as suas palavras, em inglês.
A voice pipeline can transcribe speech to text, process it, then synthesize speech. Native audio models can instead process or generate audio directly. The choice changes latency, interruption handling, and which audio information is preserved.
Vision models can consume images alongside text. For receipt extraction, evaluate resolution, rotation, missing fields, and uncertain readings before trusting structured output.
Authorization, tracing, and tool controls still apply, but preprocessing, token accounting, retrieval representations, and evaluation depend on the modality. A text-only evaluation set cannot validate image or audio behavior.
Ler é o primeiro passo. No app, essa pergunta vira flashcard com repetição espaçada, e o módulo tem quiz e simulado para testar o que ficou.
Treinar essa pergunta no app