---
title: Qwen3-TTS 1.7B Apple Silicon Benchmarks
summary: The default voice model. 4-bit, about 2.3 GB on disk and around 11 GB of memory while it runs.
---

# Qwen3-TTS 1.7B Apple Silicon Benchmarks

The default voice model. 4-bit, about 2.3 GB on disk and around 11 GB of memory while it runs.

- Repo: `mlx-community/Qwen3-TTS-12Hz-1.7B-Base-4bit`
- Maker: Alibaba
- Maker page: /resources/groundslate/benchmarks/makers/alibaba
- Line: Qwen3-TTS
- Line page: /resources/groundslate/benchmarks/makers/alibaba/qwen3-tts
- License: [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0)

Qwen3-TTS is Alibaba's speech line. The 12 Hz in every repo name is its audio token rate. GroundSlate installs four checkpoints for two jobs: a Base checkpoint speaks your scripts, and a CustomVoice or VoiceDesign checkpoint makes a voice once and is not used after that.

- GroundSlate default
- Size: 2.3 GB
- Tools: synthesize_speech, preview_voice, list_voices, get_voice, attach_voice, voice_status, list_generations
- Suite: /resources/groundslate/benchmarks/voice

## On each Mac

| Machine | Status | One line | Six lines |
| --- | --- | --- | --- |
| MacBook Pro, Apple M3 Max, 64 GB unified memory | Reference Mac | 8.57s | 14.86s |

## Speed

- One line: 8.57s, 0.95× realtime, n = 1
- Six lines: 14.86s, 1.49× realtime, n = 1