Voxtral-4B-TTS-2603-TurboQuant-MLX-8bit

8-bit MLX weight-quantized build of mistralai/Voxtral-4B-TTS-2603 with a TurboQuant KV-cache profile. Highest-fidelity MLX TTS variant on Apple Silicon — nearly indistinguishable from the reference model in subjective listening tests.

Hardware compatibility

Device	VRAM / RAM	Recommendation
Apple M4 Max 128 GB	~5.2 GB	recommended — headroom for long context
Apple M3 Max 64 GB	~5.2 GB	comfortable
Apple M2 Max 32 GB	~4.8 GB	fits

Overview

Base: mistralai/Voxtral-4B-TTS-2603 — 4B multilingual TTS with zero-shot voice cloning
Weight precision: 8-bit (group-wise)
KV-cache profile: TurboQuant
Approx. on-disk size: ~4 GB
Runtime: MLX on Apple Silicon

Quickstart

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("majentik/Voxtral-4B-TTS-2603-TurboQuant-MLX-8bit")

prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": [
        {"type": "audio", "path": "reference_voice.wav"},
        {"type": "text", "text": "Hello, this is a cloned voice."},
    ]}],
    add_generation_prompt=True,
)
audio_tokens = generate(model, tokenizer, prompt=prompt, max_tokens=2048)
# Decode acoustic tokens to waveform via the Voxtral audio decoder

Model specs

Field	Value
Parameters	4B
Weight bits	8
Group size	64
Cache profile	TurboQuant
Languages	9
Voice cloning	Zero-shot
Size on disk	~4 GB
Target hardware	Apple Silicon (M1/M2/M3/M4)
License	Apache 2.0

RotorQuant vs TurboQuant

	TurboQuant	RotorQuant
Strategy	Per-head static calibration	Rotational online re-basis
Memory reduction	~3.5x on KV-cache	~4x on KV-cache
Best for	Single-voice, single-language sessions	Multi-voice / multi-language batches

Model tree for majentik/Voxtral-4B-TTS-2603-TurboQuant-MLX-8bit

Base model

mistralai/Ministral-3-3B-Base-2512

Finetuned

mistralai/Voxtral-4B-TTS-2603

Quantized

(10)

this model

majentik
/

Voxtral-4B-TTS-2603-TurboQuant-MLX-8bit