TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Block has a page titled “Dessa: the most authentic full-duplex speech model ever built.” The supplied source material does not include technical details, evaluation results, a release date or evidence for the page’s superlative claim, so the model’s capabilities and availability remain unverified.
Block’s Dessa page describes the project as “the most authentic full-duplex speech model ever built,” presenting it as a speech technology development. But the material provided from the page is primarily a privacy notice, not a model announcement: it does not explain how Dessa works, provide test results, or say whether the model is available to use.
The page title supplies the central claim: Dessa is a full-duplex speech model and, in Block’s wording, the most authentic one built. Full-duplex generally describes systems that can handle incoming and outgoing audio at the same time, rather than requiring a strictly turn-by-turn exchange. The supplied source does not specify which capabilities Block means by “authentic,” however, or how it measured them.
The available page text instead consists of a notice about information collected when visitors use the Dessa website. It identifies Block, Inc. and its affiliates as the entities covered and lists information such as contact details users provide, device information, IP addresses, browser attributes and activity on the site. The notice says that data may be used to operate and improve the service, communicate with users, and protect the site, among other purposes.
The notice gives an effective date of October 21, 2026. It does not establish that the speech model itself launched on that date. Nor does the supplied material identify a model architecture, training approach, supported languages, latency, pricing, access conditions or a product release schedule.
Why Real-Time Speech Matters
If the description reflects a working product, a full-duplex voice system could make spoken exchanges feel less constrained by alternating turns. The potential relevance is in real-time interaction: a system able to listen while it speaks could, depending on its design, respond to interruptions or other changes in a conversation more naturally than a system that waits for a user to finish before beginning a reply.
That prospect is not a verified performance result for Dessa. The word “authentic” is a promotional claim in the page title, and the provided material contains no independent comparison, benchmark or demonstration with which to judge it. Readers and developers therefore cannot yet assess whether the model offers a practical improvement over existing voice systems, or what trade-offs it may involve.
The distinction matters because speech models can be used in customer service, assistants and other settings where interruptions, timing and conversational cues affect usability. Without information about reliability, safeguards, data handling during model use and access, those possible uses remain hypothetical rather than confirmed features of this project.
USB microphone for real-time voice AI development
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What the Dessa Page Discloses
The source is a page on block.xyz/dessa. Its supplied text begins with a privacy notice for the Dessa website and says the notice applies to information collected through that site, not to other Block products such as Square, Cash App, Afterpay or TIDAL. It also distinguishes the site’s practices from those of linked third-party services.
According to the notice, Block may collect information visitors provide directly and information generated through site use. The listed uses include providing and improving the website, analyzing usage, responding to requests, maintaining security and complying with legal requirements. The notice says information may be shared with Block affiliates and service providers. These statements concern website privacy practices; they do not describe how audio or other data would be processed by a speech model.
The source material does not include a dated technical announcement, model card, research paper, product documentation or demonstration. The headline’s description is therefore the only supplied basis for characterizing the model itself.
““This Privacy Notice applies only to information collected in connection with your access to and use of our Services.””
— Block’s privacy notice
USB audio interface for speech model testing
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Capabilities and Access Remain Unspecified
The supplied source does not confirm whether Dessa is released, available to selected users or still in development. It also gives no technical specifications, evaluation results or independent validation for the “most authentic” claim. No benchmarks, comparison methods or definition of authenticity are provided.
It is also unclear what the model can do in practice, which languages or devices it supports, whether it can be integrated by developers, and what safeguards or limitations apply. The privacy notice lists website data practices, but does not say whether model interactions involve audio recording, storage, retention or sharing. Those questions cannot be answered from the supplied page text.
closed-back headphones for voice recording and monitoring
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Awaiting Technical and Launch Details
The next useful development would be a technical announcement from Block explaining what full-duplex means for Dessa in practice, how the company evaluated the model, and whether users or developers can access it. A demonstration or documentation describing supported features, limitations and safeguards would let readers distinguish the headline claim from measured performance.
No launch date, testing schedule or further milestone appears in the supplied material. Until Block provides those details, the status and capabilities of Dessa remain unconfirmed beyond the page’s description.
Key Questions
What is Dessa?
Block’s page describes Dessa as a full-duplex speech model. The supplied source does not provide technical documentation about how it works.
What does full-duplex speech mean?
It generally refers to handling incoming and outgoing audio at the same time. The page material does not specify exactly how Dessa implements that capability.
Is Dessa available now?
The supplied material does not say whether the model is publicly available, in limited testing or not yet released.
Has Block substantiated the claim that it is the most authentic model?
Not in the provided source. The page uses that wording, but supplies no benchmark, definition of “authentic,” or independent comparison.
Does the privacy notice explain how the model handles voice data?
No. It describes information collected through the Dessa website, but the supplied text does not explain how audio from model interactions would be collected, stored or processed.
Source: hn
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
