Back to News
InnovationAI Understanding briefing

Block unveils Dessa a full-duplex speech model trained on conversation

Block's Dessa model can engage in natural-sounding conversations but its creators acknowledge it can be awkward and uncanny at times

By 4 min readRead the original reporting
Source-provided image accompanying Block unveils Dessa a full-duplex speech model trained on conversation
Attributed reportingSource recorded
Publisher
theverge.com
Source link
theverge.comhttps://www.theverge.com/ai-artificial-intelligence/1017906/jack-dorsey-block-dessa-ai-voice-speech
Source type
Reporting by a news outlet — not a first-party document.
Also cited

What we could not confirm independently: This claim is attributed to the named outlet. We did not verify it against a first-party document. (theverge.com)

Story last revised

ContextUnderstand this in 60 seconds

Start here

Key terms

AI Safety
A field focused on reducing harmful behavior, failures, and misuse risks in AI systems.
Test yourselfWhat is AI? Quiz

What changed since publication

  1. First published
  2. Block released Dessa, a full‑duplex speech‑to‑speech model trained exclusively on raw human conversation. The announcement includes live demo recordings, a public waitlist, and early access for a limited group of AI safety researchers, with claims that the model runs on consumer‑grade GPUs and captures nuanced vocal cues such as emotion and sarcasm.
  3. Block's Dessa model can engage in natural-sounding conversations but its creators acknowledge it can be awkward and uncanny at times

What happened

Block a company founded by Jack Dorsey has developed an AI model called Dessa that can engage in full-duplex speech meaning it can both listen and respond in real-time The model is trained on human conversation and can be run locally on a MacBook Pro M4 Max or better The Dessa team has shared examples of conversations with the model which can be a little awkward and uncanny at times

The Dessa model is trained on human conversation and can engage in full-duplex speech The model can be run locally on a MacBook Pro M4 Max or better The Dessa team has shared examples of conversations with the model which can be a little awkward and uncanny at times

Source details: theverge.com ↗

Why it matters

The development of Dessa is significant because it represents a step towards creating more natural and human-like interfaces between humans and AI The model's ability to learn from human speech without being trained on written text is also an important breakthrough However the team acknowledges that the model is still in its early stages and can be prone to mistakes and awkward interactions

The development of Dessa represents a step towards creating more natural and human-like interfaces between humans and AI The model's ability to learn from human speech without being trained on written text is an important breakthrough However the team acknowledges that the model is still in its early stages and can be prone to mistakes and awkward interactions The development of Dessa has the potential to improve human-AI interaction and make AI more accessible to people

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Thinking Budget (Test-Time Tokens):1,024 tokens
Complex Accuracy79%Math & Code Logic
Latency3.2sTime to first full output
Inference Cost$0.0092Per query estimated
Reasoning StyleStep VerificationInternal chain depth
Active Thinking Trace:
1Deconstruct user problem into formal constraints
2Propose candidate hypotheses & step-by-step calculation
3Self-correction: Backtrack and refute subtle edge cases
4Exhaustive consistency check & final output synthesis
Core takeaway: Test-time compute fundamentally changes AI economics. Instead of only scaling during pre-training, giving reasoning models more tokens at inference time allows them to systematically solve PhD-level STEM problems.
Interactive Concept Check+10 Points
What is AI? Quiz

A route planner searches possible journeys using explicit rules. What does this illustrate about AI?

What to watch next

The Dessa team plans to give early access to the model to a small group of researchers This will allow them to test and evaluate the model's capabilities and limitations The team's goals for the model include improving human-AI interaction and developing more natural and human-like interfaces

The Dessa team plans to give early access to the model to a small group of researchers This will allow them to test and evaluate the model's capabilities and limitations The team's goals for the model include improving human-AI interaction and developing more natural and human-like interfaces The Dessa team plans to continue refining the model and addressing its limitations

Related guides & quizzes

What is AI?AI EthicsAI AgentsAI Models ExplainedTest what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI model release tracker

Updates and corrections

This canonical story is updated in place when the developing event materially changes. Its URL and original publication date never change.

  • Block's Dessa model can engage in natural-sounding conversations but its creators acknowledge it can be awkward and uncanny at times
  • Block released Dessa, a full‑duplex speech‑to‑speech model trained exclusively on raw human conversation. The announcement includes live demo recordings, a public waitlist, and early access for a limited group of AI safety researchers, with claims that the model runs on consumer‑grade GPUs and captures nuanced vocal cues such as emotion and sarcasm.
See the public corrections log
Found this useful?