Block Dessa is the new speech-to-speech AI model from Block, the company behind Square and Cash App, and it was trained only on human conversation audio.
A very small research lab inside Block built it, and that lab reports directly to Jack Dorsey, Block's cofounder and chairman.
The model is full-duplex, so it listens and talks at the same time, and Block's page says it runs locally on a MacBook Pro M4 Max or better.
On 24 October 2026 the official page showed Download and Join the waitlist buttons, but the launch links themselves were marked as coming soon.
I haven't used Dessa myself, so this post sticks to what Block's page at block.xyz/dessa and its team's posts on X actually say.
This is the operator's briefing, which means who built it, what was announced, what you can touch today and what it changes for people who build voice agents.
I don't have a Dessa video, so here's my video on pairing Hermes Agent with ElevenLabs, which is the kind of separate voice layer Dessa is designed to do without.
The Block Dessa Briefing In One Table
This table gives you the whole announcement in one glance, so you can decide what matters for your own stack.
| Question | What Block has said |
|---|---|
| Who made it? | A very small AI research lab at Block made it, and the lab reports directly to Jack Dorsey. |
| Who are the people behind it? | Block says the lab was founded by former members of Dessa, the Toronto deep learning startup. |
| Who leads it? | Ragavan Thurairatnam signs the official page as Dessa lead. |
| What is it? | It's a full-duplex speech-to-speech model trained from scratch on human conversation audio. |
| What was it trained with? | Block says it trained Dessa on only one NVIDIA GB200 NVL72. |
| Where does it run? | The page says it runs locally on a MacBook Pro M4 Max or better. |
| What can I use now? | You can listen to the demos now, while the download and waitlist links were marked as coming soon. |
| What isn't announced? | The page gives no price, no API and no release date for the download. |
🔥 Want my exact Block Dessa setup, step by step? Inside the AI Profit Boardroom you get step-by-step video trainings, my installable Agent OS and four coaching calls a week with 3,400+ members building real automations. → Join the AI Profit Boardroom
1. Block Dessa Comes From A Tiny Lab With A Long History
Block describes the team as a very small AI research lab, and it uses the word very in brackets to make the point.
The lab was founded by former members of Dessa, a deep learning startup from Toronto, which is where the model's name comes from.
Block says the team has built deep learning models since 2012 and generative speech models for over eight years.
That history matters, because it tells you this isn't a payments company dabbling in AI for a press release.
The page also says the lab chases frontier problems others ignore and reports directly to Jack Dorsey.
The official page names six researchers, three interns and two former interns, all listed in alphabetical order.
2. Jack Dorsey Framed Block Dessa Around Voice
Jack Dorsey announced the model on X, and his post opened with the same line that tops Block's page.
He wrote, "We believe voice is the most natural way to interact with intelligence".
He also said on X that Dessa is small enough to run on a MacBook, which matches the hardware line on the official page.
For operators, the important part is the framing, because Block is betting that talking will beat typing as the way people use AI.
I've been saying the same thing about my own agents for months, because I give far more detail when I talk than when I type.
3. What Block Actually Built
Block Dessa is a true speech-to-speech model, which Block describes as audio in and audio out.
It was trained from scratch on nothing but human conversation, with no text and no pretrained models.
It's full-duplex, so it listens and talks at the same time and knows when to cut in or make small sounds that show it's following you.
Block says it natively hears and speaks emotion, sarcasm, laughter, a frustrated sigh and a voice cracking with sadness.
The page also says inference runs faster than real time, so there's no awkward pause before it answers.
Block's claim is that, as far as it knows, Dessa is the first AI you can have a realistic live conversation with that was trained only from human speech.
For the full product breakdown, read my explainer on Dessa AI, which covers every demo and what to test first.
4. Why Block Says It Built Dessa
Block gives three reasons on its page, which are curiosity, better understanding between humans and AI, and safety.
The curiosity reason is simple, because babies learn to talk without ever seeing a word, and the lab wanted to see if a model could too.
The understanding reason is that today's voice AI flattens speech into text and can't keep up with human timing.
The safety reason is the one I'd underline for any operator putting voice in front of customers.
The page says AI keeps gaining power and independence while staying nearly deaf to the emotion in our voices.
Block's example is someone in serious pain while a voice AI has no idea and just keeps going.
🔥 Want a voice agent working for you while Block Dessa is still on the waitlist? Inside the AI Profit Boardroom, I've got the Agent OS with my Hermes voice setup, plus step-by-step video tutorials on adding voice to your agents. You'll be learning alongside 3,400+ members with four coaching calls every week. → Get access here
5. What You Can Actually Touch Today
This is the part most news posts skip, so here's exactly what's open and what isn't.
| Door | Is it open? | What's behind it |
|---|---|---|
| Demo recordings | It's open now. | Six headline clips and 15 more samples, each a live, unsteered conversation. |
| Download button | It's shown, but the launch links were marked as coming soon. | The route the page shows for getting the model. |
| Join the waitlist | It's shown, but the launch links were marked as coming soon. | The other route the page shows for getting in. |
| Safety researcher access | It's open to a small group. | A form that asks for a name, an institutional email and an optional Google Scholar or blog link. |
| More recordings | It's promised. | Block says more recordings will be shared on X soon. |
The headline demos show Dessa laughing mid-sentence, getting louder when asked, finishing a forgotten film title and gasping at a surprise.
One clip has Dessa talking with GPT-Live-1, which Block identifies as ChatGPT Live, and both voices in that clip are AI-generated.
Block adds that the developer of GPT-Live-1 isn't affiliated with Block and hasn't reviewed or endorsed Dessa.
6. What Block Dessa Means For People Who Build Voice Agents
Most voice agents today are built as a chain, and mine are no exception.
Your speech becomes text, a language model writes a reply, and a voice engine such as ElevenLabs reads it back.
My Apollo copilot in the Agent OS works that way, with a wake word, speech to text, a brain that answers or hands the job to an agent, and a butler voice that replies.
Apollo also has a realtime mode where you talk continuously and can interrupt it mid-sentence.
Here's how that compares with what Block has described.
| Area | My Apollo and Hermes voice setup | Block Dessa |
|---|---|---|
| Hearing you | Speech is turned into text first. | Audio goes straight into the model. |
| Tone and emotion | Most tone is lost in the text step. | Block says it hears and speaks emotion natively. |
| Doing tasks | It opens apps, builds apps and writes notes to my vault. | The page doesn't mention tools or actions. |
| Running it | It runs inside my Agent OS on my own Mac. | The page says it runs on a MacBook Pro M4 Max or better. |
| Availability | I use it every day. | The launch links were marked as coming soon. |
My honest read is that Dessa raises the bar for how a voice should sound, while agents still own the doing.
The setup I'd want is Dessa doing the talking and an agent like Hermes doing the work.
Block hasn't published an API or developer docs, so nobody can build that combination yet.
If you're building agents now, my guide to the free Hermes Agent OS and my list of the best Hermes Agent use cases show what voice-driven agents can already do.
7. What To Watch Next From Block
These are the signals I'd track over the next few weeks.
- I'd watch for the launch links, because the Download and Join the waitlist buttons are the only routes the page shows.
- I'd watch Block's posts on X, because the page promises more recordings there soon.
- I'd watch for pricing and licence terms, because nothing about cost is published yet.
- I'd watch for an API or SDK, because that decides whether operators can build on it.
- I'd watch for findings from the safety researchers, because Block invited them in to find risks it hadn't anticipated.
If you want help planning how voice AI fits your agency or business, you can book a free SEO strategy session with my team.
The Numbers In The Block Dessa Announcement
A few numbers on the page are worth keeping in your back pocket.
Block says Dessa learned language from hundreds of times fewer words than large language models need.
The page says a single spoken word takes thousands of times more bytes than the same word in text.
Block also says it trained the model on only one NVIDIA GB200 NVL72.
The page doesn't say how many hours of audio it used, how big the model is or which languages it speaks.
I'd rather leave those blank than fill them with numbers from a summary that Block hasn't confirmed.
🔥 Want help wiring voice into your real work? The AI Profit Boardroom has the Agent OS, Hermes Agent training and four weekly coaching calls where you can ask me how to set up voice agents for your business. → Join the AI Profit Boardroom
Related Reading
- Dessa AI, the product explainer and what to test first
- The free Hermes Agent OS and what's inside it
- The best Hermes Agent use cases, ranked
Also On Our Network
- 🌐 Why a payments company like Block built Dessa, and what it means for business
- 🌐 Block Dessa against today's voice AI, with my verdict
- 🌐 Block Dessa explained for Agent OS users
- 🌐 How Block Dessa learned to talk, and why GoldieBench can't score it
FAQ: Block Dessa
What is Block Dessa?
Block Dessa is a full-duplex speech-to-speech model from a small research lab at Block, trained only on human conversation audio.
Who built Block Dessa?
A very small lab at Block built it, founded by former members of the Toronto startup Dessa and reporting directly to Jack Dorsey.
Why is it called Dessa?
The model shares its name with Dessa, the Toronto deep learning startup whose former members founded the lab.
Can I use Block Dessa today?
You can listen to the demos now, while the download and waitlist links were marked as coming soon on 24 October 2026.
Does Block Dessa replace voice agents?
Not on its own, because the page doesn't mention tools or actions, so an agent still has to do the work.
What hardware does Block Dessa run on?
Block's page says it runs locally on a MacBook Pro M4 Max or better.
If you'd like my team to look at where voice agents could fit your business, book a free SEO strategy session and we'll talk it through.
📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉
🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉
About Julian
I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
- I've built a 7-figure SEO and link-building agency, Goldie Agency, with 70+ people in Bangkok.
- I've grown a YouTube channel to 400,000+ subscribers.
- I wrote the book "Link Building Mastery", which covers nine link building strategies.
- I run GoldieBench, where I score AI models on real one-shot builds.
→ Get my best AI training inside the AI Profit Boardroom
Keep your voice agents doing the work, listen to the demos and watch for the launch links, because that's the smart way to play Block Dessa.
