Author

Andrew Zhu

@xhinker

articles: 32
total claps: 3,632
top claps: 807
avg read: 6.5 min
paywalled: 32
active: 2026-01-24 → 2026-07-17

Wiki topics

AI · AI · General · 28 LLM · Large Language Models · 19 AGT · AI Agents · 15 OPS · LLMOps & Inference · 9 💻 · Programming · 8 MM · Multimodal & Generative Media · 7 🔓 · Open Source · 6 🎮 · Gaming · 4

Top tags

#ai ·16 #ai-agent ·14 #artificial-intelligence ·14 #llm ·13 #localai ·6 #llama-cpp ·6 #local-llm ·5 #comfy-ui ·5 #stable-diffusion ·5 #programming ·5

AN Andrew Zhu · GoPenAI

Why and How to Build Your Own Local AI Machine in 2026

Everything I know how to get yourself an Local AI Machine

Mar 20, 2026 · 11 min read · 807 claps paywalled #ai #llm #ai-agent

Why and How to Build Your Own Local AI Machine in 2026

AI TCH

AN Andrew Zhu · GoPenAI

Why You Should Completely Avoid Ollama in 2026

And the way better open source options

May 23, 2026 · 6 min read · 564 claps paywalled #localai #ai #llm

AI SPT

AN Andrew Zhu · GoPenAI

I Tried Running AI Agents on My MacBook. MLX Was Too Slow. Then I Found oMLX

How a 35B parameter model went from “impossibly slow” to “production-ready” on my MacBooks — real benchmarks, real numbers, no hype.’

May 11, 2026 · 7 min read · 385 claps paywalled #omlx #ai #localai

I Tried Running AI Agents on My MacBook. MLX Was Too Slow. Then I Found oMLX

AN Andrew Zhu

DFlash Just Landed in llama.cpp: Worth to Upgrade to Get Speed Boost?

Three months ago, if you asked me what was holding back local LLM inference on consumer hardware, I’d say: too slow, especially for dense…

Jul 10, 2026 · 5 min read · 273 claps paywalled #llama-cpp #artificial-intelligence #localai

DFlash Just Landed in llama.cpp: Worth to Upgrade to Get Speed Boost?

TCH

AN Andrew Zhu · Mac O’Clock

A better way to use SSHFS in macOS Finder in 2026

Step-by-step solutions to enable macOS Finder to access and open Linux folders and files using SSHFS

Jan 26, 2026 · 3 min read · 226 claps paywalled #macos #sshfs #ssh

A better way to use SSHFS in macOS Finder in 2026

AN Andrew Zhu · GoPenAI

We Were Wrong About Anthropic

A New and Reasonable Perspective

Jun 08, 2026 · 3 min read · 184 claps paywalled #anthropics #claude-code #ai-agent

AN Andrew Zhu

You have the best image upscaler for free

A step-by-step workflow to upscale and restore image using open souce models

Jan 24, 2026 · 4 min read · 144 claps paywalled #stable-diffusion #qwen #upscale-image

You have the best image upscaler for free

AI TCH

AN Andrew Zhu · GoPenAI

ThinkingCap-Qwen3.6–27B: Think Less, Do Faster, Same Quality

Qwen3.6–27B has been my go-to model for months. It’s smart, capable, and handles everything from complex research to coding tasks. But…

Jul 15, 2026 · 4 min read · 113 claps paywalled #local-llm #ai-agent #artificial-intelligence

ThinkingCap-Qwen3.6–27B: Think Less, Do Faster, Same Quality

ECO

AN Andrew Zhu

My 5 Investment Strategies That Never Fail

My Investment Strategies: How I Never Lost a Single Dollar in the Stock Market

Jun 14, 2026 · 8 min read · 112 claps paywalled #investing #personal-finance #passive-investing

AI HUM TCH

AN Andrew Zhu · GoPenAI

What We Should Learn Next

Why to keep learning the fundamentals and coding skills — even in the age of AI

Jun 12, 2026 · 4 min read · 111 claps paywalled #ai #programming #software-engineering

AI TCH

AN Andrew Zhu

Oh, Our Safety Guardian Warns Us to Pause Frontier AI Development

Wow, wow, look at us being so responsible

Jun 06, 2026 · 7 min read · 102 claps paywalled #anthropic-claude #open-source #local-llm

AN Andrew Zhu · 𝐀𝐈 𝐦𝐨𝐧𝐤𝐬.𝐢𝐨

How to run ONE LLM model across different machines

How to pool GPUs from multiple machines into one powerful inference cluster using llama.cpp

Jun 03, 2026 · 4 min read · 100 claps paywalled #ai #localai #ai-agent

How to run ONE LLM model across different machines

AN Andrew Zhu · GoPenAI

Now NVIDIA is on Personal Local AI

What the media don’t tell you about NVIDIA GTC Taipei 2026 Keynote

Jun 01, 2026 · 3 min read · 71 claps paywalled #artificial-intelligence #nvidia #future-of-technology

AI TCH

AN Andrew Zhu · GoPenAI

Why You Should Consider Local LLM, Even have Claude Code and Codex In Hand

Because the AI trap is real

Jul 15, 2026 · 5 min read · 60 claps paywalled #artificial-intelligence #programming #open-source

AI TCH

AN Andrew Zhu

ComfyUI Scheduler vs Sampler: The Two Concepts Diffusers Hid From You

One framework calls everything “scheduler.” Another splits it into two. Understanding the difference makes you better at diffusion — no…

Jul 09, 2026 · 10 min read · 52 claps paywalled #stable-diffusion #comfy-ui #artificial-intelligence

AI MDA SCI

AN Andrew Zhu

Building Custom Audio Inference for Higgs Audio v3 TTS — A Hands-On Guide with Transformers

A practical walkthrough of how modern neural text-to-speech works under the hood — from 8 codebooks and delay patterns to autoregressive…

Jun 10, 2026 · 10 min read · 51 claps paywalled #artificial-intelligence #ai #text-to-speech

Building Custom Audio Inference for Higgs Audio v3 TTS — A Hands-On Guide with Transformers

AI MDA SCI TCH

AN Andrew Zhu · CodeX

How to Free and Reclaim VRAM in PyTorch

Every technique you need — from the basics to the nuclear option

Jun 11, 2026 · 9 min read · 51 claps paywalled #cuda #vram #python

AI GEN TCH

AN Andrew Zhu

The Rise of ComfyUI: How One Guy’s Side Project Took Over AI Image Generation

Two frameworks. Two philosophies. One won because it made impossible things possible on the hardware real people actually own

Jun 29, 2026 · 8 min read · 45 claps paywalled #comfy-ui #artificial-intelligence #open-source

The Rise of ComfyUI: How One Guy’s Side Project Took Over AI Image Generation

AI ECO MKT SPR

AN Andrew Zhu

Anthropic Discovered Nothing About AI Consciousness. They Discovered a New Way to Scare You

The art of making matrix multiplications sound like hidden thoughts

Jul 08, 2026 · 5 min read · 40 claps paywalled #artificial-intelligence #anthropic-claude #llm

AI SPT MDA

AN Andrew Zhu

One command to unlock more VRAM from Your Macbook for Free

How to Unlock MacOS RAM Limitation for LLM inference (LMStudio)

Mar 19, 2026 · 2 min read · 36 claps paywalled #ai #lm-studio #local-llm

One command to unlock more VRAM from Your Macbook for Free

AI HUM LIF MDA

AN Andrew Zhu

The Last Instruction — A Story

A short story about the one thing AI can teach us, and the one thing it can’t.

Jun 02, 2026 · 12 min read · 35 claps paywalled #artificial-intelligence #scifi #human-connection

AI MDA

AN Andrew Zhu

VoxCPM2: The TTS Model That Ditched Tokens and Changed Everything

How I spent months chasing audiobook quality — until a tokenizer-free model made the whole pipeline obsolete

Jun 18, 2026 · 5 min read · 27 claps paywalled #voxcpm #tts #ai

AN Andrew Zhu

GGUF Quantization Explained: From the Bottom Up

A practical guide to understand GGUF Quantization

Mar 16, 2026 · 5 min read · 14 claps paywalled #ai #gguf #llama-cpp

GGUF Quantization Explained: From the Bottom Up

AI MDA SPT

AN Andrew Zhu

Counter Intuitive, How Could GGUF Q8 Be Faster Than Q6 and Q4?

You’re running a large language model locally. You want speed with limited VRAM, as most of the community suggested, you pick Q4…

Jul 17, 2026 · 8 min read · 12 claps paywalled #artificial-intelligence #ai-agent #llm

Counter Intuitive, How Could GGUF Q8 Be Faster Than Q6 and Q4?

AI TCH

AN Andrew Zhu

Use ComfyUI in Pure Python Code, How To Do It Step By Step

Forget the spaghetti nodes. Control your diffusion pipeline with clean code that actually makes sense

Jul 07, 2026 · 13 min read · 5 claps paywalled #python #comfy-ui #artificial-intelligence