Former OpenAI researcher builds an AI model that judges options instead of writing text

Former OpenAI researcher Diogo Almeida has launched a model called Jev designed specifically to classify software options rather than generating text responses. By outputting structured data instead of conversational content, the model achieves response times as low as 70 milliseconds at a reduced cost. This approach aims to provide a more efficient alternative for technical applications that require binary or categorical decision-making rather than generative writing.