New on BuildTube: 2026-40

Matt Pocock’s own agent skills, straight from his .agents directory: small, composable skills for doing real engineering work with AI coding assistants rather than vibe coding.

Learn how AI agents plan, call tools, remember context and work together, one lesson at a time.

Learn to build apps with generative AI, from writing good prompts to making chat, search and image-generation apps.

Learn to build AI systems yourself, from the math behind models to retrieval pipelines and agents, by writing and running real code.

LTX-2.5 is a 22-billion-parameter open-weights model from Lightricks that generates video and sound together: describe a scene in words, or hand it a photo, and it produces a clip with a soundtrack in one pass, so picture and audio stay in sync.

Give a model up to a million tokens of text and images in one go, such as a long agent session or a large pile of input, and get text back.

Viggle’s turbo build is a distilled, faster version of Qwen-Image-2.1, the open image model: where the original takes 40 passes to render a picture, this one takes about six, which Viggle reports makes it roughly five times faster from prompt to finished image.

Unsloth’s release repackages Qwen-Image-2.1 — the open image-generation model — into GGUF files, the compact format that runs on ordinary computers through tools like llama.cpp and Stable Diffusion.cpp instead of demanding a datacenter GPU.

OpenClaw is an open-source personal AI assistant that runs on your own computer and talks to you through the apps you already use — WhatsApp, iMessage, Discord, Telegram, Slack and twenty-plus others, with native apps for every platform.

RuView reads the small disturbances people cause in the WiFi signals already filling a room and turns them into information about the space, such as whether someone is there and roughly what they are doing.

HyperFrames, from the video company HeyGen, turns a web page into a video file: you write HTML and CSS, add animations you can scrub along a timeline, and it renders a finished MP4.

OpenShell is NVIDIA’s open-source runtime for running AI agents safely: each agent gets an isolated workspace (a sandbox), and you write a policy saying what it may touch.

Let your AI assistant read web pages, YouTube subtitles, GitHub repositories, RSS feeds and, with your own logins, posts on X, Reddit and Xiaohongshu.

A skill router pack for AI coding assistants that encounter reverse-engineering tasks: an APK, a binary, encrypted frontend JavaScript, a CTF challenge, or a penetration-testing target.

Get a free domain name, such as one ending in .dpdns.org or .us.kg, and learn step by step how to point it at a website you have built.

Find language-model APIs with permanent free tiers and the page where each provider gives out keys.

A desktop app that treats agent sessions as something worth version-controlling.

An MCP server that lets AI agents and language models interact with native iOS and Android apps through one platform-agnostic interface — no separate iOS or Android expertise needed.

A modular, open-source payments infrastructure written in Rust: connect to multiple payment, payout, fraud, vault, and tokenization providers through one integration instead of one per provider.

Learn to code for free in your browser through interactive challenges, projects and exams that lead to developer certifications.

Stirling PDF is a toolbox of 50+ PDF tools: edit, merge, split, sign, redact (permanently black out text), convert, compress, and OCR (turn scanned pages into searchable text).

Build a whole web app in Rust, with pages, routing, a database and in-browser interactivity, from one framework.

Keep passwords, API keys and certificates in one encrypted store on your own servers, and control who can use each one.

Lap is an open-source, offline-first photo manager for macOS, Windows, and Linux: a privacy-focused alternative to cloud photo services with no forced upload and no subscription.

Spirula Studio trains 3D Gaussian Splatting models — from raw photos or video to splat to textured mesh — in one self-contained binary: no Python, no PyTorch, no separate COLMAP install.

A long-running open study guide for designing large-scale systems and preparing for the system design interview.

Scan or upload bills, receipts and contracts and turn them into an archive you can search by any word inside them.

Enlarge a small or pixelated photo into a bigger, cleaner image on your own computer, without uploading it anywhere.

ZDTaichu5.0-9B is a 9-billion-parameter vision-language model, meaning it reads images, video and text together.

CLM-v0.1-8B is an 8-billion-parameter model that scores choices, such as which action an AI agent should take next, instead of writing text.

Julia-1 is a 144-million-parameter decision model, tiny beside the big chat models, that takes background context, a question and a list of possible answers, then picks the right one.

Nemotron-3-Diarization is an open-weight model from NVIDIA that does speaker diarization — working out "who spoke when" in a recording — for up to eight speakers.

TeleOCR is a vision-language model of about 1.2 billion parameters (learned settings) for reading documents.