majentik/gemma-4-31B-TurboQuant-MLX-8bit

Gemma 4 31B - TurboQuant MLX 8-bit

8-bit weight-quantized MLX version of google/gemma-4-31B with TurboQuant KV-cache quantization. Optimized for Apple Silicon inference via the MLX framework.

Approximate model size: ~31 GB

Model Specifications

Property	Value
Base Model	google/gemma-4-31B
Parameters	31 billion (dense transformer)
Architecture	Dense transformer (not MoE)
Modality	Multimodal: image + text input, text output
License	Apache 2.0
Weight Quantization	8-bit (~31 GB)
KV-Cache Quantization	TurboQuant
Framework	MLX (Apple Silicon)

Quickstart

import mlx.core as mx
from mlx_lm import load, generate

model, tokenizer = load("majentik/gemma-4-31B-TurboQuant-MLX-8bit")

prompt = "Describe this image in detail."
response = generate(model, tokenizer, prompt=prompt, max_tokens=512)
print(response)

For multimodal usage with images:

from mlx_vlm import load, generate

model, processor = load("majentik/gemma-4-31B-TurboQuant-MLX-8bit")

prompt = "What do you see in this image?"
output = generate(model, processor, prompt=prompt, image="path/to/image.jpg", max_tokens=512)
print(output)

What is TurboQuant?

TurboQuant (arXiv: 2504.19874) is a KV-cache quantization technique that compresses the key-value cache used during autoregressive generation. Combined with 8-bit weight quantization in MLX, this provides a dual compression strategy: smaller model weights for reduced disk and memory footprint, plus compressed KV cache for efficient long-context generation.

KV-Cache Quantization Comparison

Method	Prefill Speed	Decode Speed	Memory Savings	Reference
TurboQuant	Baseline	Baseline	High	arXiv: 2504.19874
RotorQuant	5.3x faster	28% faster	High	GitHub

Memory Estimates (Gemma 4 31B)

Precision	Approximate Size	MLX Variant
FP16 (original)	~62 GB	--
8-bit quantized	~31 GB	This model
4-bit quantized	~17 GB	TurboQuant-MLX-4bit
2-bit quantized	~9 GB	TurboQuant-MLX-2bit

Hardware Requirements

This model requires approximately 31 GB of unified memory. Recommended hardware:

• Apple M2 Ultra (64 GB+)
• Apple M3 Max (48 GB+)
• Apple M4 Max (48 GB+)

See Also

• google/gemma-4-31B -- Base model
• majentik/gemma-4-31B-TurboQuant -- TurboQuant KV-cache only (transformers)
• majentik/gemma-4-31B-TurboQuant-MLX-4bit -- MLX 4-bit variant
• majentik/gemma-4-31B-TurboQuant-MLX-2bit -- MLX 2-bit variant
• majentik/gemma-4-31B-RotorQuant-MLX-8bit -- RotorQuant MLX 8-bit variant
• TurboQuant Paper (arXiv: 2504.19874)
• MLX Framework