Research preview — not publicly available
As of 2026-10-26 you can join Block's waitlist for Dessa, but there is no public access, no API, no pricing and no downloadable model. Early access is limited to a small group of AI safety researchers. This page tracks the status and the closest tools that work today.
Last checked:
Sources: Official Dessa page (block.xyz) · Jack Dorsey's announcement on X, 2026-10-23
Dessa AI Voice: What You Can Actually Use Today
Dessa is a full-duplex speech-to-speech model from Block, the company behind Square and Cash App, announced on October 23, 2026. Unlike a text-to-speech tool, the Dessa AI voice model listens and talks at the same time: it can be interrupted, says "yeah" and "got it" while you speak, and laughs mid-sentence.
Block says the Dessa voice model was trained only on recordings of human conversation — no text and no pretrained models — on a single NVIDIA GB200 NVL72 rack, and that it runs faster than real time on a MacBook Pro M4 Max or better. What it is not, yet, is something you can use. This page covers what exists today, what doesn't, and what to use in the meantime.
Dessa AI Voice Demo — what you can actually hear
Block published 21 recordings on the official Dessa page. We list them with the date we checked. They play on Block's site — we link to them rather than re-host them.
Dessa Alternatives — what works today
Nothing you can sign up for does exactly what Dessa does. Pick by what you actually want.
A realistic AI voice — or a clone of your own — for your content
- ElevenLabs
The most complete of the three. Instant voice cloning is included from the $6/mo Starter plan, professional cloning from Creator, and the API is documented. Start here if you need it to just work and you are not optimising for cost.
Try ElevenLabs → - Fish Audio
The cheapest of the three for volume: Plus is $11/mo billed annually for 250,000 credits, and voice cloning is listed on every plan. Free output is for personal, non-commercial use only.
Try Fish Audio → - Murf
Built around a studio timeline for voice-over rather than an API call. Voice cloning is an Enterprise add-on, so pick Murf for polished stock voices, not for cloning your own.
Try Murf →
The ElevenLabs link above is an affiliate link: if you sign up through it we may earn a commission at no extra cost to you. Fish Audio and Murf are linked without any affiliate deal. Tools are listed in the order we would recommend them.
A voice AI that listens and talks at the same time
- ChatGPT Voice (GPT-Live)
OpenAI's full-duplex voice model, live in ChatGPT since July 2026. Free users get GPT-Live-1 mini; developers can use GPT-Live-1 through the API. Block's own demos include Dessa talking to GPT-Live-1.
Open ChatGPT → - Kyutai Moshi
An open full-duplex speech model from the Paris lab Kyutai. The weights are published under CC-BY 4.0, and there is a browser demo, so you can hear the listen-while-talking style without signing up.
Try the Moshi demo →
How to Use Dessa AI
There isn't a product to walk through yet, so this is the path that exists — and the honest version of what happens at each step.
- 01
Join the waitlist
The official Dessa page has a "Join the waitlist" button. Block says launch links are "coming soon" and gives no date.
- 02
Researchers: request early access
A separate form on the same page offers early access to a small group of AI safety researchers. It asks for your name, an institutional email from a university, company or lab, and optionally a Google Scholar or blog link.
- 03
Wait for access
No public launch date has been announced. Anyone quoting one is guessing.
- 04
Use an alternative in the meantime
To hear full-duplex conversation today, try ChatGPT Voice or Kyutai's Moshi. To produce realistic voices or clone your own, use one of the three tools above.
Dessa vs ElevenLabs, Fish Audio and Murf
| Feature | Dessa AI Voice | ElevenLabs | Fish Audio | Murf |
|---|---|---|---|---|
| Public access | Waitlist only (research preview) | Open sign-up, free plan | Open sign-up, free plan | Open sign-up, free plan |
| API | None announced | Yes, documented | Yes, pay-as-you-go for paid subscribers | Yes, pay-as-you-go ($10 free trial credit) |
| Self-serve voice cloning | Not announced | Yes, from Starter | Yes, every plan | No (Enterprise add-on) |
| Commercial use | Not offered | Paid plans (Starter and up) | Paid plans (free is personal use only) | Paid plans (Creator and up) |
| Entry paid price | Not priced | $6/mo (Starter) | $11/mo billed annually (Plus) | $19/mo billed annually (Creator) |
Prices and plan features were read on each vendor's own pricing page on 2026-10-26 and change often. ElevenLabs showed a first-month promotional price on Starter that day; we list the regular price.
Dessa AI Voice FAQ
Is Dessa AI Voice available to the public?
Not yet. As of 2026-10-26 Dessa is a research preview: there is a public waitlist on Block's site, but no public access, no API and no downloadable model. We update this answer whenever the status changes.
Does Dessa have an API?
None has been announced. The official page offers a waitlist and a researcher access form — no endpoint, key or pricing page to point at.
Is Dessa's voice model free?
It isn't priced, because it isn't available. Anything you read quoting a Dessa price is speculation.
What's the difference between Dessa AI Voice and ElevenLabs?
They do different jobs. ElevenLabs turns text into speech and clones voices, and you can use it in five minutes. Dessa is a conversation model — audio in, audio out, no text step — and it isn't available yet. If your question is "which one do I use this week", the answer isn't Dessa.
Can I clone my own voice with Dessa?
Block hasn't said Dessa clones voices. It is presented as a conversation model, not a cloning tool, and there is no interface to upload a sample to. To clone your own voice today, use ElevenLabs or Fish Audio.
Is voice cloning legal in the US?
Cloning your own voice is generally fine. Using someone else's voice without permission can run into state right-of-publicity laws, and some states, such as Tennessee with its ELVIS Act, now name AI-generated voices explicitly. Get consent in writing. This is not legal advice.
How long does cloning a voice take with other tools?
Instant voice cloning — on ElevenLabs from the Starter plan and on every Fish Audio plan — works from a short sample and is meant to be usable right away. ElevenLabs' Professional Voice Cloning needs much more audio and a training period.
Will Dessa AI Voice support other languages?
Block hasn't published a language list. The official page discusses how efficiently models can learn new languages, and one recording has the model singing nonsense Spanish — neither is a supported-language claim.