Alibaba Qwen · 2025–26 · Qwen3-TTS

Qwen3-TTS

Qwen2.5-Omni and the later Qwen3-TTS line (Apache 2.0, 0.6B/1.7B) became a default local speech stack: cloning from a few seconds, streaming, natural-language direction. Sample unavailable here — weights are large.

Known for

The open all-rounder a lot of 2026 local agents actually use.

Sample unavailable. No redistributable clip was on file. Listen on the official site rather than a fake waveform.
Approach
Codec LM
Year / era
2025
License class
open
Cloning
Yes
Streaming
Yes
Emotion controls
Yes
Languages (tags)
en · zh · multilingual
Official
Source
Kokoro 82M · af_heart