#apple-silicon

LM Studio Bionic on a 64 GB M4 Max: A Fully Local AI Agent with Gemma 4 26B at 103 tok/s

LM Studio’s new agent app read my lecture notes and Jupyter notebook, wrote python-pptx code, fixed its own bugs, and delivered a…

Apple Just Turned 4,800 Hugging Face Models Into Native Mac Apps. Almost Nobody Noticed.

One free browser tool. Zero Python. Zero setup. Native Apple AI.

The Fastest AI on My Mac Isn’t Claude, GPT, or Llama. It’s This Tiny 230M Model.

The honest benchmarks, the MLX setup on Apple Silicon, and the tool-use ceiling nobody advertises

حل مشكلة تجمد النظام واختفاء الألعاب في Steam للماك M4

إذا كنت تبحث عن حل مشكلة تجميد نظام الماك M4 مع Steam واختفاء

“Apple’s AI Is Free on Linux”: A Weekend of Install Logs and One Honest Caveat.

MLX now runs on NVIDIA GPUs. Here’s what that actually buys you and what it quietly doesn’t.

حل مشكلة تجمد النظام واختفاء الألعاب في Steam للماك M4

إذا كنت تبحث عن حل مشكلة تجميد نظام الماك M4 مع Steam واختفاء

Run Qwen3–235B on a 16 GB Mac mini: 1.58-bit Ternary Quantization with TurboQuant-MLX

Per-expert disk streaming runs Qwen3–235B-A22B — a flagship open-weight MoE — on the cheapest Mac Apple sells. The headline build is a…

How ARM64 Instructions Are Really Encoded

Introduction

mlx-lm vs oMLX: I Was Wrong About the Winner

What actually separates Apple’s reference LLM runtime from the serving layer built on top of it, measured on a Mac Studio M4 Max (64GB)…

The Truth About Mac vs. NVIDIA for AI — I Tested Both

Sparse MoE vs Speculative Decoding: The Fastest Way to Run a Local Coding LLM on Apple Silicon

I benchmarked three ways to run a local agentic coding assistant on a Mac. The mixture-of-experts model reached the first token up to 6×…

Run Claude Code Locally on a Mac: 65 tok/s with a 4-bit Qwen3.6–27B and DFlash Speculative Decoding

A step-by-step guide to driving Claude Code entirely offline on Apple Silicon — no cloud, no API bill, ~19 GB of RAM, and real-time decode…

#10 The Hidden macOS That Boots When Yours Won’t

Behind your Mac’s operating system is a second, complete one whose only job is to repair the first. It has its own commands, its own…

Rosetta 2 is being phased out in macOS Golden Gate.

If you have been putting off updating a Mac because one or two older apps still rely on Rosetta, macOS Golden Gate is the version where…

Running a Local AI Development Environment on a Mac Mini M4 with Ollama

How Ollama and open-source models can reduce your dependence on Claude and GPT – without sacrificing productivity.

On Phones as Edge Computer Replacements

Why the standard “edge PC + cameras + sensors + GPIO-driven actuators” robotics stack is an artifact of when it was designed, and what a…

7 LLMs Pre-Converted to Apple’s Core AI Format (.aimodel), Now on Hugging Face

gpt-oss-20B, Qwen3, Gemma 3, Mistral — converted with Apple’s official recipes, unmodified, with hashes and measured benchmarks.

MLX Is Now a First-Class Citizen in Apple’s AI Stack: Run Any Hugging Face Model Through Foundation…

WWDC 2026 quietly did something that matters more than the Siri headlines: it opened Apple’s Foundation Models framework to almost any…

Build a Private, Offline Transcription Pipeline with Apple’s MLX

Turn meetings, interviews, and voice memos into searchable text — entirely on your Mac, for free, without sending a single byte to the…

Johny Srouji and Apple’s New Hardware Era

Apple’s design power did not disappear after Jony Ive. It moved deeper into the device.

The RTX Spark Is Not an Apple Silicon Competitor

NVIDIA’s new chip is a capable CUDA machine. Comparing it to Apple’s M-series is the wrong comparison, and in this post, I explain why.

Google Just Crushed the Memory Barrier: 32B Models Now Fit Inside 13GB

How a breakthrough quantization method from Google Research shatters the memory barrier on Apple Silicon…

Speculative Decoding on a MacBook: How MTP Landed in llama.cpp

Multi-token prediction has moved from research demos to real local inference on Apple Silicon — and the speedup you actually get depends…

 Apple Just Killed Half the Apps on Your Mac. Install These 10 Before September or Pay the Price.

macOS 26.4 started showing Rosetta deprecation warnings in February. macOS 28 ships next fall. Here’s the migration list I ran on my own…

A Sovereign Coding Agent on macOS — PI in an Apple Container; Zero npm on the Host

Run a local AI coding agent that reads code and runs shell commands — without handing it your filesystem, your credentials, or a Node…

Why Apple Silicon Inference Needs Diffusion Drafters, Not Diffusion LLMs

Two years into the diffusion-LLM experiment, the most interesting thing diffusion can do for language models is not generate them.

From Swift to Silicon: A Beginner’s Roadmap to Metal Programming with C++14 on Apple Silicon

A primer for iOS and macOS developers stepping into GPU programming for the first time, structured around the upcoming book “Metal…

Llama.cpp vs MLX on Apple Mx

Why MLX Has Fewer Parameters Than llama.cpp, and Why That Doesn’t Make It Worse

Running Gemma 4 Locally with MLX

In my previous post, I explored what it looks like to run coding agents with local LLMs on an M3 Ultra Mac Studio. That write-up was mostly…

How A-Series Apple Silicon Powers My Daily Compute

My computing needs vary throughout the day. On non-work days, I mostly use my iPhone 17 Pro because it’s super convenient and always with…

I Benchmarked Qwen3.6–35B-A3B Model on 3090, 4090, 5090 and M5 Max. Here’s What Nobody Tells You.

RTX 3090, 4090, 5090, and Apple Silicon (M5 Max) — same model, same engine, brutally honest results.

Run OpenCode with Gemma 4 Locally on Apple Silicon

A complete step-by-step guide to setting up a private, zero-cost AI coding agent on your Mac. No API keys, no cloud, no data leaving your…

How My Local Coding Agent Crashed My Mac, and What I Learned About MLX Memory Management

Made with AI

Why My MacBook Air M4 Runs an 8B Model at 45 Tokens per Second

How Apple silicon’s unified memory and parallel GPU execution make local inference feel like a perfectly coordinated burger line with help…

How I run 122B-parameter LLMs on a MacBook — outperforming MXFP4 and standard quantization on Apple…

I extended TurboQuant to Mixture-of-Experts models. A 120B LLM that no existing format could fit on 64GB now runs at 44 tokens per second…

From a Commodore 64 to Gaussian Splatting — Why I Built RadianceKit for the Mac

I’ve been making 3D things on home computers since the 80s. This is the latest one.

Why Your M4 Max Runs LLMs Like a GPU Beast — Without a Single CUDA Core

The secret isn’t brute-force compute. It’s a memory architecture that rewrites the rules of local AI inference.

Apple Accelerates Its Unified Silicon Strategy

Apple’s latest Mac announcements may look like routine upgrades. Look closer, and a deliberate strategy behind the new MacBook Neo and…

oMLX 效能調校 KV Cache 與Concurrent Batching

上一篇 oMLX vs Ollama Mac 本地推論Qwen3.5–35B實測 踹完之後，一直很好奇介紹中提到的 KV Cache 與 Concurrent Batching 跟過去理解的有什麼差異？

Apple M5: Il silicio ha tre anime

Apple M5: Il silicio ha tre anime

Running AirLLM Locally on Apple Silicon: Not So Good

This week, armed with an article on huggingface talking about how AirLLM can run 70b models on 4GB of GPU, I thought. my M4 MacBook Pro…

Fixing Screen Size Issues in Kali Linux on VirtualBox (Apple Silicon)

Hi. If you’re running Kali Linux on VirtualBox on an Apple Silicon Mac (M1, M2, M3), you’ll likely hit an annoying problem:

Fixing Sliver “Exec format error” on Kali Linux ARM (Parallels / Apple Silicon)

If you’re running Kali Linux ARM (aarch64) on Apple Silicon (M1/M2/M3) using Parallels and try to install Sliver C2 using the official…

How I Got Age of Empires II: DE Running Smoothly on an M3 Mac for FREE (Using VMware Fusion +…

Got Windows 11 ARM and Age of Empires II running smoothly on an M3 Mac using VMware Fusion and free tools. Here’s exactly how I did it.

How To Setup Linux on Apple M-Chips

Run Linux on Apple M-series Macs easily! Learn how to set up Linux using QEMU, UTM, and OrbStack — fast, reliable, and ARM-optimized.

🚀 Supercharge Ollama on Your M1/M2 Mac with Apple Metal Performance Shaders (MPS)

3–5× faster AI inference, less battery drain, and zero cloud costs

Apple Silicon: R, VS code and Quarto

This blog post focuses on setting up Visual Studio Code (VS Code) for R programming on MacBooks with Apple Silicon chips (M1, M2, etc.). If…

Echoes of 1997: What Steve Jobs’ WWDC Fireside Chat Can Teach Today’s Leaders in 2025

YouTube is a wonderful tool. I frequently stumble across a video that makes me pause and think about things. This happened to me a few days…

How to install Keil C51 on MacOS with Apple Silicon C

Step 1: Install wine-stable via homebrew package manager.

OrbStack: lightweight Linux VM for Apple Silicon

OrbStack shared kernel approach brings WSL 2 to macOS