LaunchAI ModelsNovember 10, 2026

Zyphra releases ZONOS3-Mini open-source TTS model

Read original source →marktechpost.com

Sparse MoE with 800M total and 100M active parameters (64 experts, top-2 routing), Apache 2.0 weights, zero-shot voice cloning and word-level delivery control. Zyphra reports 80 ms median time to first audio on one H100 at 8 concurrent requests and up to 10 minutes of audio per generation.

1 source

More stories today

Open the live feed