Skip to main content

VeloxQuant-MLXGetting Started Algorithms Guides API Reference Changelog Blog

All posts

2026

The ROM Chip That Wasn't, and the 230× Speedup That Was
The Sign Was the Whole Paper: Debugging a KV Cache Compressor on Real Models
A 5.65× Metal Kernel
TensorOps Research: What We Learned Optimizing KV Caches
KIVI: The Most-Cited KV Cache Baseline, Implemented in MLX
Hands-On: Compressing Your First LLM with VeloxQuant-MLX
I Wrote a Metal Kernel to Stop My Mac From OOMing
TurboQuant + Metal Kernels: The Combined Writeup
Benchmark Results: 10 Models, 8 Compression Configs
I Ported a Google Research Paper to Apple Silicon
VeloxQuant-MLX: Fast KV Cache Quantization for Apple Silicon

Tags

A

apple-silicon11

B

benchmarking2
benchmarks3

G

gpu4

K

kivi2
kv-cache6

M

memory1
metal5
mlx11

P

performance3

Q

q-filters1
quantization4

R

research1

T

turboquant1
tutorial1

W

weights1

Docs

Getting Started
Algorithms
Guides
API Reference

Community

GitHub
Issues
☕ Buy me a coffee

More

Blog
Changelog
Home

MIT License © 2026 VeloxQuant-MLX. Built with Docusaurus.