Sign in

Nick

@w512.bsky.social
20 followers 3 following 60 posts

Software engineer 👨‍💻 ( JavaScript 🌐 | Vue.js 🧪 | Python 🐍 | Swift 🐦‍⬛ ) youtube.com/@ZeroToMVP

PostsRepliesMedia
Nick @w512.bsky.social · 29/09/2026
Just because a model runs on a Raspberry Pi doesn't mean it's usable. 8 models. 2 criteria. 4 tasks. 3 made it. youtu.be/5bubV4v79uU #RaspberryPi #LocalLLM
youtu.be
8 Local AI Models on a Raspberry Pi 5: Only 3 Are Usable
YouTube video by Zero to MVP
001
Nick @w512.bsky.social · 25/09/2026
GPT-6 Sol vs Claude Opus 5.5 🥊 Same 3 tasks, each model in its native tool: Codex vs Claude Code. An evolution sim, a real code review, and more. One came out ahead, but it was closer than I expected. youtu.be/7N82XCJJXLs #GPT6 #ClaudeOpus #AICoding
youtu.be
GPT-6 Sol vs Claude Opus 5.5: Which New Model Should You Use?
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 22/09/2026
Four local LLMs, one 16GB GPU. Chat, coding agent, document indexer, OCR — all fighting for the same card. So I built a scheduler that decides who gets it. Chat always wins; background tasks pause and resume on their own. New video 👇 youtu.be/Mu3xzCVoHXc #LocalLLM #LocalAI
youtu.be
Run Multiple Local LLMs on One GPU Without Slowdowns
YouTube video by Zero to MVP
110
Nick @w512.bsky.social · 17/09/2026
Do you need a powerful desktop to run local AI? I ran 6 LLMs on a regular home computer. Qwen 3.8 27B fits and runs at ~14 tok/s. Here are the two models I'm keeping on it 👇 youtu.be/jaHVQnJAnSw #LocalAI #Ollama #Qwen
youtu.be
What Local AI Models Can a Regular Computer Run?
YouTube video by Zero to MVP
021
Nick @w512.bsky.social · 05/09/2026
Finally open-sourced my coding agent: Green Agent. Inspired by Pi agent, plus a full-screen mode (project tree + live file viewer), a browser for docs, and approvals that only ask when git can't undo it. Works with any OpenAI-compatible API. MIT. github.com/w512/green-a...
220
Nick @w512.bsky.social · 21/08/2026
Ornith 1.5 dropped two days ago, so I ran both sizes on my MacBook. 9B dense — fine. 35B is 23GB on a 24GB machine and still loaded — it's MoE, ~3B active per token. Not a review. Just first run + real speed numbers on M4 Pro. youtu.be/fP9hBrweli4 #LocalLLM #AppleSilicon
youtu.be
Ornith 1.5 Speed on a MacBook M4 Pro — 9B and 35B
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 20/08/2026
Ran Qwen 3.8 27B through three coding projects locally, from an HTML clock to a full-stack app. It wrote every piece of code I asked for. It couldn't design the architecture — a bigger model found 6 critical flaws in its plan. #LocalLLM #Qwen #Ollama youtu.be/dTutfoSVMq4
youtu.be
Qwen 3.8 Locally: It Can Code, But It Can't Plan
YouTube video by Zero to MVP
100
Nick @w512.bsky.social · 13/08/2026
Small LLMs aren't weaker versions of big models — they're a different tool. New video: 4 ways I use 2–4 GB models running 24/7 on a low-power home server. OCR, summaries, translation, and a private medical assistant. youtu.be/9LkxI2H3gI0 #LocalAI #SmallLanguageModels #HomeLab
youtu.be
4 Ways I Use Tiny LLMs on My Always-On Home Server
YouTube video by Zero to MVP
100
Nick @w512.bsky.social · 29/07/2026
Can a 12B local model build real apps? Yes — but only if you do half the work. I spent 3 hours with Gemma 4 12B Coder. It finished the project, but needed the plan broken into 18 tiny steps and constant supervision. Where exactly it breaks 👇 youtu.be/HpLFqf8sIPI
youtu.be
Gemma 4 12B Coder Tested: Where Small Local Models Break
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 23/07/2026
Qwen 3.6 built me a local OCR app last video. Qwen 3.8 just finished it. Four features, no implementation plan, no subtasks — just the requirement in a file and a "go." Repo is tagged at the handoff point, so you can see which model wrote what. youtu.be/1RA9w2IrJJI #Qwen #LocalLLM #AICoding
youtu.be
Qwen 3.8 Finished the App Qwen 3.6 Started
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 16/07/2026
Everyone's hyped about Fable 5 and GPT-5.6. But can a small LLM running on YOUR computer actually build a useful app? I spent 4-5 hours finding out — Qwen 3.6 + Pi coding agent, zero cloud. The result surprised me 👇 youtu.be/WzCk5G_gGTE #LocalLLM #Qwen #AICoding
youtu.be
Can a Small Local LLM Actually Build a Useful App?
YouTube video by Zero to MVP
101
Nick @w512.bsky.social · 21/06/2026
I launched AI Vault: an awesome-style list for developers building with AI or learning AI engineering. I’m trying to keep the list useful, not turn it into a graveyard of random links. GitHub: github.com/w512/AI-Vault Stars, PRs, and tool recommendations are very welcome.
000
Nick @w512.bsky.social · 19/06/2026
GLM 5.2 vs Kimi 2.7 Coder — two of the strongest Chinese models, head-to-head on real coding tasks. Running both in the cloud via @ollama + Pi agent. The result was closer than I expected. youtu.be/mnivzCQ9Y1Q #LocalAI #AICoding #LLM #GLM52 #Kimi27
youtu.be
GLM 5.2 vs Kimi K2.7 Code — Which Is Better for Coding?
YouTube video by Zero to MVP
030
Nick @w512.bsky.social · 17/06/2026
Z.AI says GLM 5.2 beats GPT 5.5. So I put GLM 5.2 through 3 escalating coding tasks: easy sort algo → six-algorithm visualizer → Rust epidemic sim. Zero hand-holding. Here's how it did 👇 youtu.be/OgRp6lDuzTY #GLM52 #LocalAI
youtu.be
Testing GLM 5.2 on Easy, Medium, and Hard Coding Tasks
YouTube video by Zero to MVP
111
Nick @w512.bsky.social · 16/06/2026
New video: setting up Hermes Agent end to end — installing it, hooking up a model (local or API), and running your first session. youtu.be/Gs1J1SGCW_k #HermesAgent #AIAgent #AgenticAI
youtu.be
How to Install & Set Up Hermes Agent (Step-by-Step Guide)
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 14/06/2026
I started a newsletter. Zero to MVP Weekly: one AI signal, one tool, and one MVP idea you could actually build — every Saturday, in about 5 minutes. Not a news digest. A filter for people who build. Free → weekly.blokhin.us
weekly.blokhin.us
Zero to MVP Weekly
AI tools, coding agents, and MVP ideas — filtered for builders.
000
Nick @w512.bsky.social · 10/06/2026
Anthropic's new Fable 5 burns 2x the tokens of Opus 4.8. So I ran them head-to-head: same Rust task, same prompts, two folders. Plan it, build it, compare. Is the upgrade worth double the cost? My honest take 👇 youtu.be/y7WhesEusmM #ClaudeCode #Fable5 #AICoding #Rust
youtu.be
Claude Fable 5 Uses 2x the Tokens — Is It Worth It?
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 01/06/2026
Can an AI agent orchestrate Claude Code for you? Hermes can. I walk through the setup, the two run modes, and my honest take on whether it beats just using Claude Code directly. New video out now 👇 youtu.be/0p3eyo_nfTg #ClaudeCode #AIagents #DevTools
youtu.be
Can Hermes Agent Use Claude Code? (Yes — Here's How)
YouTube video by Zero to MVP
130
Nick @w512.bsky.social · 25/05/2026
New experiment: multi-model coding with Hermes on a ZimaBoard 2. - Hermes orchestrates - DeepSeek 4 plans + verifies - MiniMax 2.7 writes code - Two real projects built almost autonomously Watch how far it actually got 👇 youtu.be/-3MPnUGqa68 #LocalAI #LLM #AIAgents #ZimaBoard
youtu.be
Hermes + DeepSeek 4 + MiniMax 2.7: Multi-Model Coding on a ZimaBoard
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 20/05/2026
Gemini 3.5 Flash dropped yesterday, so I tested it the way I actually use these models: building a real app. Two prompts. One implementation plan. One full build. The result genuinely surprised me. New video on the channel 👇 youtu.be/zaO588Ic4kQ #Gemini35Flash #Antigravity #LocalLLM #AICoding
youtu.be
Testing Gemini 3.5 Flash: Building a LLM Benchmark App
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 17/05/2026
I took the Markdown editor that Gemma 4 built in my last video and refined it into a real, open-source app. Free, open source, macOS + Windows builds available 👇 youtu.be/vh5LyL8Y164 #Texodus #Gemma4 #LocalLLM #Tauri #VueJS #OpenSource #MarkdownEditor
youtu.be
From Gemma 4 Test Task to Real App: Introducing Texodus
YouTube video by Zero to MVP
050
Nick @w512.bsky.social · 13/05/2026
New video is up 🎥 I gave Qwen 3.6 (27B) and Gemma 4 (31B) the exact same task: build a cross-platform markdown editor with Tauri. Same prompt, same hardware, one winner? Spoiler: the timing difference surprised me. youtu.be/Um8Px55mINc #LocalLLM #Qwen #Gemma #AICoding #SelfHosted
youtu.be
Qwen 3.6 vs Gemma 4: I Built the Same App With Both Locally
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 07/05/2026
Can a locally-running LLM handle my daily coding tasks? I tested Qwen 3.6 27B on three real projects — from a simple to-do app to a full Kanban board with drag-and-drop. The result surprised me. New video 👇 youtu.be/7ejQSGaiyQE #Qwen #LocalLLM #AICoding #SelfHosted
youtu.be
Testing Qwen 3.6 Locally on Easy, Medium, and Hard Coding Tasks
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 01/05/2026
New video is up! 🎬 How to use local LLMs in Zed Editor with Ollama and LM Studio — including running models on another machine in your local network. Private, free, and no API keys needed. youtu.be/Bq2po2T3h7k #ZedEditor #LocalLLM #Ollama #LMStudio #AICoding
youtu.be
Zed Editor + Local LLMs: Full Setup Guide with Ollama and LM Studio
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 26/04/2026
Just dropped: How to connect DeepSeek v4 to VS Code in under 5 minutes — no Copilot subscription needed. 🧵👇 youtu.be/6OWz9sT0QYY #DeepSeek #VSCode #AICoding #DeveloperTools #Programming
youtu.be
How to Connect DeepSeek 4 to VS Code in 2 Minutes
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 25/04/2026
Can DeepSeek 4 really compete with Claude and Gemini? I put it to the test on a ZimaBoard 2 with Hermes Agent — building a sorting visualizer and a full ToDo app from scratch. Spoiler: I'm impressed. Watch here 👇 youtu.be/5SAjq9VLl5w #DeepSeek #DeepSeek4 #HermesAgent #ZimaBoard #ZimaBoard2
youtu.be
Testing DeepSeek 4 + Hermes Agent on ZimaBoard 2
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 22/04/2026
New video: Hermes Agent on a Raspberry Pi — full setup in under 8 minutes. A self-improving AI agent that remembers decisions, creates skills, and reuses past experience. youtu.be/dduq8mA8Vxs #HermesAgent #RaspberryPi #AIAgent #SelfHostedAI #Homelab
youtu.be
Hermes Agent on Raspberry Pi: Self-Improving AI in Under 8 Minutes
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 19/04/2026
Tested the new Claude Opus 4.7 on an actual project instead of benchmarks. Stack: Tauri + Vue + Rust Goal: desktop client for Ollama & LM Studio Tool: Claude Code, "xhigh" effort Full breakdown in the video 👇 youtu.be/HbM1Qu9a0Z8 #ClaudeOpus #ClaudeCode #Tauri #Ollama #LocalLLM
youtu.be
Claude Opus 4.7 First Impressions: A Real-World Coding Test
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 17/04/2026
New video is up! 🎥 I show how to run Claude Code directly inside VS Code — and then connect a local LLM (Gemma, Qwen, DeepSeek…) to the same setup. All in a few minutes. 👇 youtu.be/GImq1WL9OJQ #ClaudeCode #VSCode #LocalLLM #AI #Coding
youtu.be
How I Use Claude Code with VS Code (and Local Models)
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 13/04/2026
Just dropped a new video on how to run local AI models directly inside VS Code — no Cursor, no Copilot, no API keys needed. Using Gemma 4 + Ollama. Full setup from scratch 👇 youtu.be/89bhDV0FBSM #VSCode #LocalAI #Ollama #Gemma4 #OpenSource
youtu.be
How I Use Local AI Models in VS Code (Free & No API Key)
YouTube video by Zero to MVP
151
Nick @w512.bsky.social · 09/04/2026
Just dropped a new video: how to use Claude Code with a local Gemma 4 model — no API costs, runs on your own machine. Works with Qwen and other local LLMs too 👇 youtu.be/sKNq4CqWkT4 #ClaudeCode #Gemma4 #LocalLLM #LMStudio #AIcoding #GenerativeAI #OpenSourceAI #DevTools #MachineLearning #AIdev
youtu.be
Claude Code with Gemma 4 (How I Use It)
YouTube video by Zero to MVP
460
Nick @w512.bsky.social · 05/04/2026
New video: Running Gemma 4 on Raspberry Pi 5 with LM Studio CLI, exposing it over local network, and connecting from Zed Editor. Here's the full walkthrough 👇 youtu.be/kZhAj8--t8w #RaspberryPi #RaspberryPi5 #Gemma4 #LocalAI #LLM
youtu.be
Running Gemma 4 on a Raspberry Pi 5 — Can It Actually Work?
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 04/04/2026
Google just dropped Gemma 4 and it’s a game changer for the local LLM community! 🚀 I tested the new models on my MacBook and Desktop. From vision tasks to complex coding, does it actually live up to the hype? Watch here: youtu.be/T6AvsQVSL74 #Gemma4 #GoogleAI #LocalLLM
youtu.be
Gemma 4 - I Tested it on My Laptop and Desktop
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 01/04/2026
New video: Running Nemotron 3 Super (120B params, 83GB) locally with LM Studio on an RTX 4060 Ti. Mixture of Experts makes it possible. ~6.5 tok/sec on consumer hardware. Full walkthrough + settings 👇 youtu.be/advIzSzMAew #LocalLLM #AI #Nemotron #LMStudio #OpenSourceAI #GPU
youtu.be
Nemotron 3 Super: Can an 83GB Model Run on 16GB VRAM?
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 27/03/2026
I tested Qwen Coder Next — a free, local coding model that runs entirely on your own hardware. No tokens, no monthly fees. Here's what it can (and can't) do 👇 youtu.be/jDeeoHSc2kw #AI #LocalLLM #QwenCoder #CodingTools
youtu.be
Qwen Coder Next Locally: Can It Replace Paid AI Models?
YouTube video by Zero to MVP
020
Nick @w512.bsky.social · 20/03/2026
I compiled Hello World in Rust, Nim & C and compared the binary sizes. Rust fans… you might want to sit down for this one 😅 Know a better way to shrink a Rust binary? Drop it below 👇 🎥 youtu.be/-MCJ-V-6r9E #rust #rustlang #c #nim #nimlang
youtu.be
Rust vs Nim vs C: How Big Is Your Hello World?
YouTube video by Zero to MVP
100
Nick @w512.bsky.social · 18/03/2026
Tired of hitting token limits on ChatGPT and Claude? 😤 I tested Qwen 3.5 — a FREE open-source AI you can run locally on your own machine. The results were... surprising. 👀 Full video here 👇 youtu.be/FcR-gUGfS4E #AI #LocalAI #Qwen #OpenSource #LLM
youtu.be
Qwen 3.5 Locally: Can It Replace Paid AI Models?
YouTube video by Zero to MVP
120
Nick @w512.bsky.social · 02/03/2026
After 15 years of using password managers — 1Password, Enpass, Strongbox — I finally said “enough” and built my own. Not because they’re bad. But because none of them was exactly what I wanted. New video: why I made this decision and how it all started 👇 youtu.be/FOmuucZ4dfE
youtu.be
Why I Decided to Build My Own Password Manager
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 20/02/2026
Gemini 3.1 might be the best AI coding assistant right now. I tested it. Two tasks. Zero manual fixes. Both worked on the first try. Judge for yourself 👇 youtu.be/9r57r2iIjKM #Gemini3 #AITools #CodingWithAI
youtu.be
Gemini 3.1 Is Here — I Tested It So You Don't Have To
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 14/02/2026
After a year with the same theme, I finally decided to switch things up 🎨 Tested 8+ color themes for Zed Editor to find the perfect one. Which one won? 👇 youtu.be/54SUJFU8Lwo #ZedEditor #DevTools #CodingLife
youtu.be
Choosing a New Color Theme for Zed Editor
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 31/01/2026
Tired of AI subscription fees? I just made the simplest tutorial on running AI models locally - for FREE. Just 3 steps. No cloud. No costs. No limits. Watch: youtu.be/XMSROaxZ4hU
youtu.be
Run AI Models Locally for FREE - Simple 3-Step Guide (Ollama Tutorial)
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 25/01/2026
7 MCP servers that make AI actually useful (I use them every single day): 1. Context 7 2. GitHub 3. Slack 4. PostgreSQL 5. Linear 6. Playwright 7. Brave Search Full breakdown 👇 youtu.be/b1J0jaS6mTQ
youtu.be
7 MCP Servers That Turn AI Into a Real Developer Assistant
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 22/01/2026
I've been using MCP for almost a year now. It completely transformed how I work with AI. No more copy-paste. No more outdated docs. Just seamless integration. Here's everything I wish I knew when I started: youtu.be/aLG81S__uys
youtu.be
How MCP Turns AI from Chatbot into Real Assistant
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 07/01/2026
Just published a new video on backing up GitHub repositories 🔒 Here are 3 methods to properly backup your repos (including private ones): youtu.be/Jx4JS7z8OKU
youtu.be
3 Ways to Backup All Your GitHub Repositories
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 23/12/2025
The end of the year is time for backups : )
000
Nick @w512.bsky.social · 19/12/2025
Can AI actually control your browser? 🤖 I tested Google Antigravity's new browser integration with Gemini 3 Pro. The results surprised me: www.youtube.com/watch?v=v0xq...
youtube.com
I Tested Google Antigravity’s Browser Control — How It Works
YouTube video by Zero to MVP
000
Nick @w512.bsky.social · 16/12/2025
#morning
000
Nick @w512.bsky.social · 13/12/2025
Hot take: Google Antigravity is better than Cursor. Multiple simultaneous agents > single agent with better UI #AI #programming #google #Antigravity youtu.be/gpifYYje0jQ
youtu.be
Google Antigravity Explained: Complete Beginner Tutorial to Get Started Fast
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 03/12/2025
If you're trying out the Zed editor, you might be looking for that "Live Server" experience. I broke down the best methods to get it running smoothly. #Zed #WebDevelopment #DevTips #ZedEditor #SoftwareEngineering 👇 youtu.be/R_2GXd0Qc_Y
youtu.be
Zed Live Server: The BEST Way to Run Live Reload in Zed
YouTube video by Zero to MVP
010
Nick @w512.bsky.social · 03/12/2025
Which one should I choose for work today? : ) #mechanical #keyboard #keychron #nuphy
020