{"id":"zonos3-mini","slug":"zonos3-mini","name":"ZONOS3-Mini","orgId":"zyphra","family":"zonos3-mini","releaseDate":null,"availability":"ga","openWeights":true,"license":"Model weights Apache License 2.0 (blog and Hugging Face card); inference runtime repository MIT (GitHub Zyphra/ZONOS3-Mini LICENSE)","paramsTotal":null,"paramsSource":null,"contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://www.zyphra.com/our-work/zonos3-mini","hfRepo":null,"modalities":["text","audio"],"aliases":["ZONOS3-Mini","zonos3-mini","zonos 3 mini","zonos3 mini","Zonos 3 Mini","zyphra zonos3-mini","Zyphra/ZONOS3-Mini"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-11-10","announced":"2026-11-09","blurb":"Zyphra's open text-to-speech model with zero-shot voice cloning and word-level control of emotion, pace and delivery. A sparse MoE with 800 million total parameters and 100 million active, licensed Apache 2.0. It runs on-device (browser WebGPU, phones, native GGML/MLX) and on servers (SGLang, vLLM), and is also served on Zyphra Inference / Zyphra Cloud.","api":"live","openWeights":"yes","params":{"text":"800M total / 100M active (sparse MoE)","source":"Zyphra ZONOS3-Mini blog and Hugging Face model card"},"facts":[{"label":"Type","value":"Zero-shot voice-cloning text-to-speech; 11-layer hybrid of Gated DeltaNet and global attention (3:1) with latent MoE (Zyphra blog and GitHub README)"},{"label":"Codec","value":"Zyphra Audio Codec 1 (ZAC-1). GitHub README: the model predicts 10 × 4096 dual-codec audio codes at 15 Hz; a causal multi-resolution iSTFT decoder yields 44.1 kHz audio"},{"label":"Voice control","value":"Identity (age, gender, accent, quality) via speaker embeddings and accent task vectors; delivery (pace, emotion, projection, liveliness) via activation steering, including word-level steering from alignment attention heads, plus a post-generation edit model"},{"label":"Speaker conditioning","value":"Two modes: a clip in context (pretrained; more natural, does not copy prosody exactly) or a direct voice representation of the target clip (short finetune after pretraining). Speaker encoder is MECT-B2 (Zyphra blog)"},{"label":"Training","value":"112 NVIDIA H100s in 2.5 days, then DPO and GRPO post-training (Zyphra blog)"},{"label":"Runtimes","value":"Day-0 SGLang (blog also lists vLLM); native CUDA, MLX, C++/GGML; in-browser WebGPU. 27 default voices ship in the Hugging Face repo"},{"label":"Hosted access","value":"Zyphra Inference on NVIDIA H200s and Zyphra Cloud (zyphra.com / cloud.zyphra.com)"},{"label":"Weights","value":"Public, ungated Hugging Face repo Zyphra/ZONOS3-Mini (PyTorch, GGUF, WebGPU shards)"},{"label":"License","value":"Model weights Apache License 2.0 (blog and Hugging Face card); inference runtime repository MIT (GitHub Zyphra/ZONOS3-Mini LICENSE)"},{"label":"Code","value":"https://github.com/Zyphra/ZONOS3-Mini"},{"label":"Related","value":"Separate from earlier Zyphra Zonos-v0.1, ZONOS2 and Zamba models; those are not this page"}],"claims":["Lab claim: runs in real time with under 570 ms time-to-first audio in the browser on a Pixel 9 and under 150 ms on an M3 MacBook Air (Zyphra blog).","Lab claim: under similar data constraints, a model trained with ZAC-1 has a 10x lower word error rate than one trained with Descript Audio Codec, with better acoustic similarity (Zyphra blog).","Lab claim: self-reported on Zyphra's own ZTTS2-Eval / TTS Capability Index v1 (420 items), ZONOS3-Mini ranks first among open models, followed by Fish S2 Pro and MOSS TTS v1.5."],"sources":[{"kind":"lab","title":"ZONOS3-Mini blog (Zyphra)","url":"https://www.zyphra.com/our-work/zonos3-mini","date":"2026-11-09"},{"kind":"lab","title":"Zyphra on X: ZONOS3-Mini","url":"https://x.com/ZyphraAI/status/2119914895861076151","date":"2026-11-09"},{"kind":"lab","title":"Zyphra on X: ZONOS3-Mini thread (weights, blog, code)","url":"https://x.com/ZyphraAI/status/2119914911837061605","date":"2026-11-09"},{"kind":"repo","title":"Zyphra/ZONOS3-Mini on Hugging Face","url":"https://huggingface.co/Zyphra/ZONOS3-Mini","date":"2026-11-09"},{"kind":"repo","title":"Zyphra/ZONOS3-Mini on GitHub","url":"https://github.com/Zyphra/ZONOS3-Mini"}],"firstBenchmark":null},"organization":{"id":"zyphra","name":"Zyphra","slug":"zyphra","country":null,"url":"https://www.zyphra.com","color":"#11110F","mark":"Z"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Nov 9, 2026; no independent results yet.","lead":"As of Nov 22, 2026, ZONOS3-Mini is pending benchmarks on isitsota: announced Nov 9, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}