Dessa

Block

2026-10-29 08:39:29

Block unveils Dessa, a full-duplex voice model trained only on speech

Block’s AI research lab has introduced Dessa, a full-duplex voice model trained exclusively on human conversational speech data. The company said the model does not rely on text data or any pretrained model, setting it apart from systems that use text as an intermediate layer. Dessa uses a speech-to-speech architecture, allowing it to listen and speak at the same time. According to the announcement cited by Techub News, the model supports more natural interactions such as real-time interruption and laughing in the middle of a sentence. For training, Dessa was built on a single NVIDIA GB200 NVL72 rack equipped with 72 GPUs. On the inference side, Block said the model can run locally on a MacBook Pro M4 Max or higher-spec devices, with performance faster than real time. The model is still in a research preview stage. Block has not released model weights, a paper, an API, or pricing information.

260

Jack Dorsey

2026-10-26 09:14:31

Dorsey says there is no moat, Saylor replies that Bitcoin is the moat

Techub News reported that Jack Dorsey posted on X saying there is "no moat," and the post drew more than 3.5 million views. Michael Saylor answered with a short rebuttal, writing that "Bitcoin is the moat." Elon Musk joined the exchange as well, posting that "Sparta never had a moat." The discussion came shortly after Block introduced its voice AI model, Dessa. According to Block, the model was trained only on human conversations, without text data and without any pretrained model. The company also said training can be done on a single Nvidia GB200 NVL72 rack. The item cited BeInCrypto as the source for the underlying report.

60

Block

2026-10-24 15:21:06

Block unveils Dessa voice AI, drawing praise for humanlike speech and warnings over scam risks

Block, the payments company led by Jack Dorsey, has introduced a voice AI model called Dessa that the company says was trained entirely on human speech rather than text. In Block’s demonstrations, the system speaks with pauses, laughter, tonal shifts and interruptions that made some viewers say it sounded unusually close to a real person. Dorsey said talking with it felt like talking to a person, while social media reactions ranged from admiration to concern. According to Block, Dessa processes audio directly instead of converting speech into text first and then generating a spoken reply. The company said it trained the model from scratch on a single Nvidia GB200 NVL72 system and that it is compact enough to run locally on a MacBook Pro with an M4 Max chip or better. For now, though, Dessa is only a research preview. Block has not disclosed the model’s size, benchmark results or a public release date. The demo also triggered debate about misuse. Binance founder Changpeng Zhao pointed to possible education use cases through his GiggleAcademy project, while filmmaker Justine Bateman and journalist Shiv Aroor warned that highly humanlike AI voices could make deception and scam calls harder to detect. Block said early access is being given to a small group of AI safety researchers as those risks are assessed.

140