Dessa is a full-duplex speech-to-speech model from a small AI lab inside Block, released on October 23, 2026 as a research preview. It takes audio in and returns audio out. Block says the model was trained from scratch on recordings of human conversation, with no text and no pretrained model underneath. Early access is limited to a small group of AI safety researchers, who apply with a university, company, or lab email. A public waitlist is open for everyone else, and Block has not published weights, an API, or a price. The model listens and speaks at the same time. Block says it learned from real conversation when to cut in and when to answer with short backchannels such as "yeah" or "got it," and that it hears and produces emotion, sarcasm, laughter, sighs, and a voice cracking with sadness. The company says inference runs faster than real time, that Dessa runs locally on a MacBook Pro with an M4 Max or better, and that training fit on a single NVIDIA GB200 NVL72. It also says the model learned language from hundreds of times fewer words than large language models. No parameter count, dataset size, latency figure, or formal benchmark has been published. What Block has shared instead is a set of recordings it describes as live and unsteered, with subtitles added afterward. The lab was started by former members of the Toronto deep learning startup Dessa, is led by Ragavan Thurairatnam, and reports to Block cofounder Jack Dorsey. The team says it has built deep learning models since 2012 and generative speech models for more than eight years. Block, the payments company behind Square and Cash App, frames the work around curiosity, a more natural interface between people and AI, and a safety concern that voice systems which turn speech into text cannot hear emotion.