Dessa: first full-duplex speech-to-speech model trained entirely on speech, runs on consumer GPUs
ZeroStateReflex · x · 2026-10-24
The Dessa team announced what they call the first AI model for realistic conversation trained entirely on speech: a full-duplex speech-to-speech model that runs on consumer GPUs. They call it 'a GPT-3 moment for speech.'
Unlike cascaded ASR→LLM→TTS stacks, end-to-end full-duplex models allow natural turn-taking and instant response, with local deployment lowering the barrier.
Related event: Block's Dessa: Voice-Trained Full-Duplex Speech Model Runs on a MacBook(8 posts)→
More from Models
- Kimi K3.1 falls short of benchmarks in hands-on testing, says KOL — DeryaTR_ · 2026-10-24
- Kimi K3.1 Pro tops agentic coding benchmarks but may lag on AA-3D — ArtificialAnlys · 2026-10-24
- ValsAI Index Finds Fireworks AI Beats DeepSeek and Moonshot's Own APIs on Agentic Tasks — sophiamyang · 2026-10-24
- TensorSharp hits 5.5× llama.cpp prefill throughput on DeepSeek V4.1 Flash across five A40s — fuzhongkai · 2026-10-24
- Kimi weights rumored to land Nov 4: would you self-host a 2.8T MoE at ~$2/$10 API pricing? — AIFlow_ML · 2026-10-24
- GLM shines at network/ops/security tasks despite trailing frontier models, dev says — davidcrawshaw · 2026-10-24