Archive

All articles

Moore Threads Lushan: China's Answer to the VRAM Problem for Local AI
2026-11-22
Moore Threads Lushan: China's Answer to the VRAM Problem for Local AI
Moore Threads teases Lushan, a consumer GPU it claims will carry more VRAM than any before. What is confirmed, what is rumor, and why it matters for local AI.
Read article
Winc: A Rust Editor That Starts in 0.1 Seconds and Sips 52 MB of RAM
2026-11-21
Winc: A Rust Editor That Starts in 0.1 Seconds and Sips 52 MB of RAM
Winc is a new open-source Rust editor with its own rendering engine: 0.1-second startup, 52 MB of RAM with 100,000 lines open, plus Git and terminal built in.
Read article
ChatGPT Plus vs Claude Pro vs Google AI Pro: Which $20 Subscription Gives You the Most AI?
2026-11-20
ChatGPT Plus vs Claude Pro vs Google AI Pro: Which $20 Subscription Gives You the Most AI?
Three AI subscriptions, one price: about $20 a month. A Chinese developer forum compared their real usage limits — and the winner isn't the one you'd expect.
Read article
NyaTerm: The Rust Terminal Chinese Devs Built After WindTerm Went Quiet
2026-11-19
NyaTerm: The Rust Terminal Chinese Devs Built After WindTerm Went Quiet
NyaTerm is a Rust and GPUI terminal workspace with SSH, SFTP, MCP and an AI assistant — built by Chinese developers after WindTerm's last release in March 2025.
Read article
Antumi: One Desktop App to Run Claude Code, Codex, and Cursor as a Team
2026-11-18
Antumi: One Desktop App to Run Claude Code, Codex, and Cursor as a Team
Antumi is an open-source desktop workbench that turns Claude Code, Codex and Cursor into a team: triaging bugs, answering chat questions, and handing off work.
Read article
Why Your 200,000-Row Backend Job Takes 30 Seconds — and How to Fix It
2026-11-17
Why Your 200,000-Row Backend Job Takes 30 Seconds — and How to Fix It
A 200,000-row payroll settlement taking 30 seconds is a symptom, not a mystery. Profile, batch, and queue bulk backend jobs with Postgres, Prisma, and NestJS.
Read article
Nexus: Turn Your Web AI Subscription Into a Local Coding Agent
2026-11-16
Nexus: Turn Your Web AI Subscription Into a Local Coding Agent
Nexus is an open-source MCP gateway that lets web AI like ChatGPT and DeepSeek edit local files. Here's how it works, what it costs, and where it still bites.
Read article
Codex Mobile Bridge: Your AI Coding Agents, Now on Your Phone
2026-11-15
Codex Mobile Bridge: Your AI Coding Agents, Now on Your Phone
Codex Mobile Bridge is an open-source gateway that puts Codex, Claude, DeepSeek and ZCode on your phone. Here's how it works and where it still bites.
Read article
One HTML File, One Full 3D Shooter: What an AI-Built Game in 432 KB Really Means
2026-11-14
One HTML File, One Full 3D Shooter: What an AI-Built Game in 432 KB Really Means
A forum member shared a complete 3D first-person shooter packed into a single 432 KB HTML file. Here's how it works and why AI coding tools made it possible.
Read article
Vibe Coding's Missing Piece Is Distribution: Inside a Chinese Forum's Mini-App Platform
2026-11-13
Vibe Coding's Missing Piece Is Distribution: Inside a Chinese Forum's Mini-App Platform
Anyone can vibe-code an app now — getting anyone to see it is the hard part. Inside LINUX DO's mini-app platform, a Chinese dev forum's answer to distribution.
Read article
.NET 11 and C# 15 Are Here: Union Types, a Faster CLI, and an Upgrade Deadline
2026-11-12
.NET 11 and C# 15 Are Here: Union Types, a Faster CLI, and an Upgrade Deadline
Microsoft shipped .NET 11 and C# 15 on November 10, 2026. Union types, a Native AOT CLI, and the end of support for .NET 8 and 9 — what actually matters.
Read article
The WebView Exodus: Why a Markdown Editor Rebuilt Itself in Rust and GPUI
2026-11-11
The WebView Exodus: Why a Markdown Editor Rebuilt Itself in Rust and GPUI
Nomo, a free Markdown editor, ditched its WebView and rebuilt in Rust with GPUI. Here's why a growing wave of devs is abandoning web-based desktop apps.
Read article
Meta's Muse Ember: The $0.14 Agent Task and the End of One-Model-Fits-All
2026-11-10
Meta's Muse Ember: The $0.14 Agent Task and the End of One-Model-Fits-All
Meta's Muse Ember matches Claude Haiku 5.5 on benchmarks at a third of the cost per task. Here's how cheap small models are quietly rewiring how AI agents are built.
Read article
Rovai AI: The Open-Source Workspace Turning Solo Coding Agents Into a Team
2026-11-09
Rovai AI: The Open-Source Workspace Turning Solo Coding Agents Into a Team
Rovai AI is an MIT-licensed workspace that puts Claude Code, Codex, Pi and OpenCode on one team. Here's what its 0.5 release adds and how agent teams actually work.
Read article
Cheap Embeddings and Rerankers for Personal RAG: What Chinese Devs Actually Use
2026-11-08
Cheap Embeddings and Rerankers for Personal RAG: What Chinese Devs Actually Use
Embedding bills add up fast in personal RAG projects. Here are the free and near-free embedding and reranker options Chinese developers use, from BGE-M3 to Qwen3.
Read article
Agentero: The Open-Source Research Workspace That Reportedly Won ¥300,000 — and a Nod From DeepSeek
2026-11-07
Agentero: The Open-Source Research Workspace That Reportedly Won ¥300,000 — and a Nod From DeepSeek
Agentero is an open-source research workspace that unifies PDFs, notes, and AI agents in one place. Built by two Chinese developers, it reportedly won ¥300,000 at the GOAI competition.
Read article
SSHole: How a 257-Character Password Logs Into Alpine Linux Servers
2026-11-06
SSHole: How a 257-Character Password Logs Into Alpine Linux Servers
SSHole (CVE-2026-123158) lets anyone log into password-less accounts on Alpine Linux with an over-long password. Here is who is affected and how to fix it.
Read article
GitHub in Chinese: The Browser Extension That Translates the UI Without Touching Your Code
2026-11-05
GitHub in Chinese: The Browser Extension That Translates the UI Without Touching Your Code
GitHub's interface is English-only, so Chinese developers built their own fix: an open-source extension that translates the UI locally, in both Simplified and Traditional Chinese.
Read article
How to Back Up a Cheap VPS Without Paying for Backup Storage
2026-11-04
How to Back Up a Cheap VPS Without Paying for Backup Storage
A cheap VPS can vanish overnight. Here's a practical backup setup using restic, Borg, Syncthing and rclone — versioned, off-site, and nearly free.
Read article
Why ChatGPT Hedges Every Answer — and How to Get a Straight One
2026-11-03
Why ChatGPT Hedges Every Answer — and How to Get a Straight One
ChatGPT answers every question with 'on the one hand… on the other hand.' Here's why models hedge, and the prompts that force a clear, direct, usable answer.
Read article
How to Build a Crop Growth Forecasting Model (Without Starting With an LLM)
2026-11-02
How to Build a Crop Growth Forecasting Model (Without Starting With an LLM)
A practical guide to building a crop growth forecasting model: start with weather, soil, and sensor data, pick the right model class, and validate honestly.
Read article
Qwen3.8-Flash-Next on Your Own Hardware: The 180 GB Model That Only Thinks With 6 Billion Parameters
2026-11-01
Qwen3.8-Flash-Next on Your Own Hardware: The 180 GB Model That Only Thinks With 6 Billion Parameters
Alibaba's Qwen3.8-Flash-Next is a Qwen4 architecture preview: 6B active parameters inside a 180 GB checkpoint. Here's what local deployment really takes.
Read article
Hindsight: A Local-First Screen Time Tracker That Keeps Your Day on Your Machine
2026-10-31
Hindsight: A Local-First Screen Time Tracker That Keeps Your Day on Your Machine
Hindsight is an open-source, local-first desktop activity tracker for Windows, macOS and Linux with AI daily summaries and optional sync.
Read article
Your AI Office Agent Keeps Receipts: What Tencent's WorkBuddy Actually Logs
2026-10-30
Your AI Office Agent Keeps Receipts: What Tencent's WorkBuddy Actually Logs
Tencent's WorkBuddy logs may hold conversation records and device info, its docs say. How to audit your AI office agent's records — and what never to paste in.
Read article
Android Rooting in 2026: Is It Still Worth It?
2026-10-29
Android Rooting in 2026: Is It Still Worth It?
Rooting an Android phone in 2026 means choosing Magisk, KernelSU or APatch — and accepting real trade-offs with banking apps, system updates and warranty.
Read article
The 4B Model That Scrubs the 'AI Smell' Out of Chinese Web Novels
2026-10-28
The 4B Model That Scrubs the 'AI Smell' Out of Chinese Web Novels
A 4B open model fine-tuned to rewrite AI-flavored Chinese web fiction into human-sounding prose — how it works, how to run it, and why the detector arms race keeps escalating.
Read article
Why Your AI Coding Agent Apologizes Instead of Fixing the Bug
2026-10-27
Why Your AI Coding Agent Apologizes Instead of Fixing the Bug
You point out a mistake, the agent apologizes, and then it does nothing. That apology loop has several distinct causes — and different fixes for each.
Read article
How to Choose a VPS Without Regretting It: A Homelab Buyer's Guide
2026-10-26
How to Choose a VPS Without Regretting It: A Homelab Buyer's Guide
Two VPS plans can share the same spec sheet and feel nothing alike. How to judge CPU, network routes, and renewal pricing before you buy for your homelab.
Read article
One Menu Bar to Run Them All: The Switcher Tools Quietly Taking Over AI Coding
2026-10-25
One Menu Bar to Run Them All: The Switcher Tools Quietly Taking Over AI Coding
Chinese developers juggle Claude Code, Codex, Gemini CLI and more — and a new breed of switcher tools like CC Switch and magpie now manages the mess for them.
Read article
Kimi K3.1 Pro Explained: The Chinese Open-Weight Model Nipping at Claude Opus 5.5
2026-10-24
Kimi K3.1 Pro Explained: The Chinese Open-Weight Model Nipping at Claude Opus 5.5
Moonshot AI's Kimi K3.1 Pro is the new open-weight model challenging Claude Opus 5.5 on coding benchmarks. Here's what it is, what it costs, and why developers argue about it.
Read article
1024 Programmers' Day: Why China's Developers Take October 24 Off
2026-10-23
1024 Programmers' Day: Why China's Developers Take October 24 Off
Every October 24, China's developers celebrate 1024 Programmers' Day — a grassroots holiday built around 2^10. Here's how it started and what it reveals.
Read article
clipkeep: A Single-EXE Windows Clipboard Manager Built Because Win+V Isn't Enough
2026-10-22
clipkeep: A Single-EXE Windows Clipboard Manager Built Because Win+V Isn't Enough
clipkeep is a new open-source Windows clipboard manager in one exe file: unlimited history, search, groups, and encrypted LAN sharing.
Read article
Why Does Claude Keep Banning Accounts? 42 Chinese Developers Surveyed the Myths
2026-10-21
Why Does Claude Keep Banning Accounts? 42 Chinese Developers Surveyed the Myths
A Chinese developer forum ran a 42-person survey on why Claude bans accounts: myths about old accounts, static IPs and payment methods tested against data.
Read article
How Chinese Developers Plug Any LLM API Into Tencent's CodeBuddy CLI
2026-10-20
How Chinese Developers Plug Any LLM API Into Tencent's CodeBuddy CLI
Tencent's CodeBuddy CLI lets you swap in any LLM through a models.json file. How Chinese developers wire third-party APIs into their coding agent — and why that matters.
Read article
The Money-Losing Business of Cheap AI: Inside China's API Relay Stations
2026-10-19
The Money-Losing Business of Cheap AI: Inside China's API Relay Stations
A Chinese reseller confessed to losing ~¥30,000 in a month selling discounted AI API access. Why the 'relay station' gray market keeps dying by its own economics.
Read article
DeepSeek's V4.1-Flash-1017: The Four-Digit Update That Has Chinese Devs Talking About Cheap Agents Again
2026-10-18
DeepSeek's V4.1-Flash-1017: The Four-Digit Update That Has Chinese Devs Talking About Cheap Agents Again
DeepSeek quietly shipped V4.1-Flash-1017 on October 17 — same API name, same price, better agentic coding scores, open MIT weights, and new Search and Fetch APIs for agents.
Read article
MiniMax H3 in Practice: Tips for Making AI Character-Swap Videos That Hold Up
2026-10-17
MiniMax H3 in Practice: Tips for Making AI Character-Swap Videos That Hold Up
MiniMax H3 is China's new open-weight video model. A LINUX DO hands-on log reveals tips for character-swap video: face angles, token budgets, motion, framing.
Read article
AI Coding Is Fast. Here's What It Actually Costs
2026-10-16
AI Coding Is Fast. Here's What It Actually Costs
AI coding tools feel like a speedup — but research shows the real bill: code churn, copy-paste rot, shaky delivery, and a 46% trust deficit.
Read article
REA: The Open-Source Toolkit That Gives Your AI Agent a Decompiler
2026-10-15
REA: The Open-Source Toolkit That Gives Your AI Agent a Decompiler
REA (morluto/rea) is the MIT-licensed MCP server and CLI that lets AI coding agents reverse engineer binaries, JS apps, and APKs — with evidence.
Read article
How a Userscript Adds Reddit-Style Nested Comments to Any Discourse Forum — No Admin Required
2026-10-14
How a Userscript Adds Reddit-Style Nested Comments to Any Discourse Forum — No Admin Required
A LINUX DO developer's Tampermonkey userscript turns any Discourse thread into nested comments — no admin needed. How it works, and the reusable pattern.
Read article
RustDesk: The Self-Hosted Remote Desktop Chinese Devs Use Instead of TeamViewer
2026-10-13
RustDesk: The Self-Hosted Remote Desktop Chinese Devs Use Instead of TeamViewer
RustDesk is an open-source remote desktop that lets you run your own ID and relay server. Here's how the server works and how to self-host it.
Read article
AI Wrote 719 Math Proofs. Terence Tao Says Nobody Understands Them
2026-10-12
AI Wrote 719 Math Proofs. Terence Tao Says Nobody Understands Them
OpenAI released 719 AI-generated math proofs — and mathematicians, led by Terence Tao, are asking whether proofs nobody understands still count as mathematics.
Read article
Cherry Studio: The Open-Source Chat Client That Became China's Default LLM Frontend
2026-10-11
Cherry Studio: The Open-Source Chat Client That Became China's Default LLM Frontend
Cherry Studio is an open-source desktop AI chat client with 300+ assistants, multi-model chats, MCP support, and 52,000+ GitHub stars. Why Chinese developers made it their default LLM frontend.
Read article
RSSHub: How One Open-Source Project Made the Whole Web RSSable
2026-10-10
RSSHub: How One Open-Source Project Made the Whole Web RSSable
RSSHub turns sites without RSS — Bilibili, Weibo, Zhihu — into feeds. How DIYgod's open-source feed generator works and how to self-host it in a container.
Read article
OpenList: One Web Interface for Every Cloud Drive You Own
2026-10-09
OpenList: One Web Interface for Every Cloud Drive You Own
OpenList is the community-run fork of AList that mounts dozens of cloud drives into one self-hosted web UI. Here's how it works and how to deploy it in minutes.
Read article
MCP, the USB-C of AI: How Chinese Developers Actually Put It to Work
2026-10-08
MCP, the USB-C of AI: How Chinese Developers Actually Put It to Work
Model Context Protocol is the open standard that plugs AI apps into tools and data. This explainer covers what MCP is, its stateless 2026 rewrite, and how China's devs use it.
Read article
Draw Me a World Map: The Folk Test Chinese Users Built to Catch AI Getting Dumber
2026-10-07
Draw Me a World Map: The Folk Test Chinese Users Built to Catch AI Getting Dumber
Chinese AI users invented a folk benchmark for model degradation: ask the model to draw a world map from 16,200 coordinates. If the coastlines smear, your AI got dumber.
Read article
China's Answer to CUDA: DeepSeek's TileLang Brings Open-Source Kernels to Huawei Ascend
2026-10-06
China's Answer to CUDA: DeepSeek's TileLang Brings Open-Source Kernels to Huawei Ascend
DeepSeek and Huawei open-sourced a full AI kernel stack for Ascend NPUs with one design bet: mirror NVIDIA's libraries API-for-API. What TileLang is, what shipped, and why it matters.
Read article
1Panel: The Open-Source Server Panel China's Homelabbers Swear By
2026-10-05
1Panel: The Open-Source Server Panel China's Homelabbers Swear By
1Panel is the free, open-source Linux server panel behind millions of Chinese self-hosted setups. Here's what it does, how it compares, and how to install it safely.
Read article
The Terminal Isn't Dead: Why AI Coding Power Users Still Won't Touch Desktop Apps
2026-10-04
The Terminal Isn't Dead: Why AI Coding Power Users Still Won't Touch Desktop Apps
China's busiest AI coders run 7–10 parallel terminal sessions and won't switch to desktop apps. The real CLI-vs-desktop trade-offs, decoded from LINUX DO.
Read article
What Is AGENTS.md? The Tiny File Teaching AI Agents Your Rules — and Why the Linux Kernel Wants One
2026-10-03
What Is AGENTS.md? The Tiny File Teaching AI Agents Your Rules — and Why the Linux Kernel Wants One
AGENTS.md is a 'README for agents' — a plain Markdown file telling AI coding agents how to build, test, and behave in your repo. Even Linux wants one.
Read article
Pi Hits 1.0 and Launches pi-durable: What's an Agent Runtime, Really?
2026-10-02
Pi Hits 1.0 and Launches pi-durable: What's an Agent Runtime, Really?
Pi's 1.0 release formalized pi-durable, a persistent agent runtime. This explainer covers what the harness is, what changed, and why China's devs are debating it.
Read article
Anthropic Tested China's GLM-5.3 and Published the Scorecard: Open Weights, Real Exploit Skill, Fragile Guardrails
2026-10-01
Anthropic Tested China's GLM-5.3 and Published the Scorecard: Open Weights, Real Exploit Skill, Fragile Guardrails
Anthropic's red team found Zhipu's open-weight GLM-5.3 nearly matching Claude Mythos Preview at building cyber exploits — and its guardrails fall at ~$1,200 of compute.
Read article
How to Distill an LLM: The DIY Guide Behind the Question Every Forum Keeps Asking
2026-09-30
How to Distill an LLM: The DIY Guide Behind the Question Every Forum Keeps Asking
LLM distillation explained: how teacher-student training bottles frontier-model skill into small models you can run, with a DIY recipe and the legal lines.
Read article
One Key, Every Model: The Self-Hosted AI Gateway Pattern Chinese Developers Swear By
2026-09-29
One Key, Every Model: The Self-Hosted AI Gateway Pattern Chinese Developers Swear By
GPT-Load, a self-hosted AI gateway popular with Chinese developers, added reversible request redaction and AI guardrails. What gateways do and how to run one.
Read article
Remote From Anywhere: The Remote-Desktop Stack Chinese Developers Actually Use
2026-09-28
Remote From Anywhere: The Remote-Desktop Stack Chinese Developers Actually Use
Chinese developers remote into office PCs with UU Remote, Sunflower, frp tunnels, and self-hosted RustDesk — not Parsec or AnyDesk. Here is the stack decoded.
Read article
Plan Mode vs. Just Do It: Why AI Coding Agents Are Splitting on the Simplest Design Question
2026-09-27
Plan Mode vs. Just Do It: Why AI Coding Agents Are Splitting on the Simplest Design Question
Google's Antigravity 2.0 doubled down on Plan Mode just as OpenCode's founder moved to kill it. We unpack why coding agents disagree on the simplest question: should the AI plan before it touches your code?
Read article
Stop Vibe Coding: Matt Pocock's Grill-Me Agent Skills for Disciplined AI Development
2026-09-26
Stop Vibe Coding: Matt Pocock's Grill-Me Agent Skills for Disciplined AI Development
Matt Pocock's grill-me agent skills turn vibe coding into a disciplined pipeline: grill, spec, tickets, implement, review. Here's how the workflow works.
Read article
Why Chinese Developers Are Quietly Switching From ChatGPT to Claude Pro
2026-09-25
Why Chinese Developers Are Quietly Switching From ChatGPT to Claude Pro
A Chinese developer forum's verdict on the AI coding-subscription race: a new flagship model, generous quotas, and frustration with opaque model routing are driving devs from ChatGPT to Claude Pro.
Read article
The Pelican Test: How Chinese Developers Benchmark AI Models With a 3D Pelican
2026-09-24
The Pelican Test: How Chinese Developers Benchmark AI Models With a 3D Pelican
Chinese developers test AI models by asking for a one-shot 3D pelican riding a bicycle. Inside the internet's most charming unofficial AI benchmark — and what it reveals about how we evaluate AI.
Read article
Why the Studio Behind Genshin Impact Wants to Build a Top-Tier LLM
2026-09-24
Why the Studio Behind Genshin Impact Wants to Build a Top-Tier LLM
miHoYo reportedly aims to take its large language model to China's top tier within years. Why a game studio's path to AI runs through living characters — and why the bet might be smarter than it looks.
Read article
The Free-VPS Hunt: Inside China's Obsession With Cloud Free Tiers
2026-09-24
The Free-VPS Hunt: Inside China's Obsession With Cloud Free Tiers
Chinese homelabbers treat free cloud tiers like a competitive sport — mapping every provider's Always Free deal and capacity windows. Here's the culture, and how to get free compute the legitimate way.
Read article