VOXTRAL: MISTRAL'S OPEN-SOURCE AUDIO MODEL

Summary from the TLDR AI newsletter

Mistral's has released Voxtral, its first open-source audio model suite. It features a 24B parameter model for large-scale use and a 3B variant for edge deployment.