298 subscribers
14K videos
14.6K links
Download Telegram
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ geramy/lemonseed-studio

Code on iPad with a GPU

Run a full development environment on an iPad, powered by a separate graphics card you plug in via cable. LemonSeed Studio is a single app that acts as an editor and terminal, but its secret sauce is how it connects to hardware. It embeds a specialized driver that lets the tablet talk directly to an external processor, bypassing the usual limits of mobile devices. This means you can run complex language models right on that attached hardware, keeping your work private and your code flowing without waiting for a cloud server. It is a clever workaround that turns a tablet into a serious workstation.

πŸ“° https://news.ycombinator.com/item?id=50041573

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ grimthorpe/clbre

Clbre: The AI-Free Calibre

Clbre is a fork of the popular Calibre ebook manager that has completely stripped out all artificial intelligence features. While the main project now includes AI-powered tools, this version removes them to give you a leaner, traditional experience without the extra overhead. It keeps the essential functions you expect, like organizing your library, editing metadata, and converting between different ebook formats. By cutting the AI, the project aims to offer a cleaner, faster alternative for those who prefer a straightforward tool without the complexity of modern machine learning integrations. It is a clear statement of preference for simplicity and direct control over your digital reading collection.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ shauryagangrade/awesome-ai-prompts

The Anti-Hallucination Prompt Pack

Awesome AI Prompts is the prompt collection that forces your coding agent to actually prove its work instead of just guessing. Most AI tools will confidently invent fake solutions that look perfect but break your code, so this project solves that by giving you copy-paste instructions that demand evidence. You paste one of these prompts into your agent, and it is strictly forbidden from claiming success until it has run the tests and verified the results against your actual files. It turns your AI assistant from a creative writer into a disciplined senior engineer who checks its own homework.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ taoliveaigc/taomate-fvsr

TaoMate: Face Video Restoration

Bring blurry face videos back to life by letting the audio guide the visual repair. TaoMate FVSR takes a low-quality clip and reconstructs the face using the sound track to ensure the mouth movements match perfectly, solving the problem of unnatural lip sync in restored footage. It works by analyzing the speech to drive the facial animation, creating a result that looks sharp and natural. This is useful for anyone needing to fix old recordings or create high-quality talking head videos without manual editing. You can run it locally or plug it into ComfyUI for a visual workflow.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ jorjbarbosa/naruto-rise-recomp

Naruto: Rise of a Ninja Recomp

Naruto: Rise of a Ninja Recomp is the PC port that brings this Xbox 360 classic to modern screens by rebuilding its original code from the ground up. Instead of running the old console hardware through a slow emulator, this project translates the game's internal instructions into fresh native code that runs directly on your computer, making the entire experience feel brand new. It solves the problem of outdated hardware limits by letting you play with modern graphics settings, including ultrawide display support and high resolution scaling. This means you can enjoy a familiar story with crisp visuals and smooth performance on your current setup.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ zortax/zgui

zgui: The Rust UI Framework That Thinks Like a Browser

zgui lets you build native desktop apps in Rust that actually feel alive, ditching the web browser entirely. It’s a retained mode framework, meaning your code runs once, but tiny signals update only the exact pixels that changed. No full redraws, no JavaScript, just pure GPU power. It even uses real CSS for styling, so your familiar web skills finally work on the desktop. If you want snappy, high-performance interfaces without the bloat, this is the tool you need to try.

πŸ“° https://news.ycombinator.com/item?id=50373205

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ iyilang/iyi

iyi

iyi lets you delete the actual source code for your dependencies and still build your program successfully. This is a new programming language designed to be incredibly fast for both human developers and AI agents. It works by compiling your code against just the interface declarations, not the heavy implementation details. Because of this, your builds are tiny, your startup time is milliseconds, and AI tools can analyze your project without needing to read the entire repository. It also runs natively on many different operating systems.

πŸ“° https://news.ycombinator.com/item?id=50405667

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ omnijev/flide

Flide: 9.5x Faster Decision Inference

Flide answers multiple questions about the same situation in a single pass, making it up to 9.5 times faster than standard inference engines. It uses exact state forking to split one context into several questions without repeating work, so every answer is exactly what the model would give if asked alone. This infrastructure is built for decision models that need to make fast, accurate choices on real data. It solves the problem of slow, redundant processing when an AI has to evaluate several criteria at once. The result is a tool that delivers precise, independent answers at high speed, which is critical for real-time applications.

πŸ“° https://news.ycombinator.com/item?id=50421155

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ master-cai/cola3d

CoLa3D: One Image, Many Objects

One photo turns into a fully interactive 3D scene where every single object is a separate, movable piece. That is the entire trick of CoLa3D, a research project accepted into NeurIPS twenty twenty-six. It solves a massive headache for creators who need to edit digital environments. Instead of giving you one solid block that you have to manually chop apart with digital scissors, the software automatically recognizes the sofa, the lamp, and the table as distinct entities within a shared space. You can prompt it to isolate specific items by pointing, masking, or letting it figure it out on its own.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ artcraft-android/procraft

ProCraft: 3D Painting on Your Tablet

Paint three dimensional art directly on your Android tablet using ProCraft, an open source app that brings professional digital painting to touch screens. The project stands out because it lets you paint textures onto 3D models using pressure sensitive brushes and layers, all running on a custom Rust engine that handles the graphics directly on your device. This means you get a responsive creative tool without relying on heavy external software, making it perfect for artists who want a portable, high performance studio in their pocket. If you are looking for a way to combine digital illustration with 3D design, this is a fascinating open source project to explore.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ marvis-ai/marvis

Marvis: The AI That Never Leaves Your Desk

Keep your AI assistant strictly on your own computer. Marvis is a desktop app that floats above your workspace as a small, translucent bar. It watches your screen and listens to your meetings to answer questions, all without sending your data to a cloud server. This means your API keys and screen history stay safe on your device. The most interesting part is that it hears both your microphone and your system audio, creating a live transcript with speaker labels so you can see exactly who said what. It feels like having a private, intelligent secretary who never leaves your office.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ zju3dv/splatanything

SplatAnything: 3D from Any Camera

SplatAnything takes a handful of photos and spits out a full, navigable 3D scene. Most tools need you to calibrate cameras first, but this model works from raw images alone, making it perfect for turning a single 360-degree panorama into a realistic virtual room or reconstructing a street from a car’s dashcam. It’s the rare AI that handles everything from a tiny teacup to an entire city skyline without needing a new setup each time. If you want to build 3D assets without a laser scanner, this is the one to try.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ lins-lab/ultratext_bench

UltraText Bench: The Truth About AI Text

UltraText Bench puts image generators on trial to see if they can actually write real words. It tests if models can render dense English and Chinese text correctly across entire scenes, checking everything from tiny labels to massive posters. The coolest part is that it catches when text looks sharp but says the wrong thing. It uses an automated judge to score accuracy, clarity, and placement, giving you a clear number on whether an AI is truly reliable for text. So the next time an image looks perfect, remember that looking good is not the same as being right, and this tool tells you the difference.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ vectorspacelab/arex-2

AREX-2: The AI That Gets Smarter As It Works

Watch an AI agent get genuinely better at a task just by trying again. AREX-2 is a research project that tests if large language models can learn to fix their own mistakes in real-time, round after round, without needing a full retraining. The key finding is that this self-correction skill, built on coding and math problems, actually transfers to deep research tasks like searching the web for answers. This repo gives you the exact tools to run those tests yourself, ensuring every score is tied to a specific model version and data snapshot so you can trust the results.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ qlabs-eng/noise

Noise: Pretraining Without Backprop

A team built a speedrun where the only rule is you cannot use backpropagation. They froze a thirty-seven million parameter transformer model and a fixed dataset, then challenged researchers to train it using only forward passes. No gradients, no autograd, just pure estimation. The current best score beats a standard training run, proving that better search algorithms are finally catching up to the default. It is a direct test of whether brute force learning can actually replace the core engine of deep learning. If you care about the future of AI, this is where the experiment is happening. Watch this space, because the records are already breaking.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ apexdesk/apexdesk

ApexDesk: The Office Suite In One File

ApexDesk is a full office suite that fits inside a single executable file, with zero dependencies. No installation, no cloud account, and no telemetry; it just runs offline on your machine. Written in Rust, it compiles to one tiny native binary that opens instantly, handling documents, spreadsheets, and presentations just like the heavy desktop apps you are used to. Because it is a standalone program, you can drop it on any Windows, Linux, or Android device and start working immediately. It even works in the terminal for quick edits. So if you are done waiting for Office to load, this is the lightweight alternative that actually works.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ 0xbakeer/tandemllm

Run a 27 Billion Parameter AI Model Locally on One Tiny Box

Turn a single compact desktop into a private AI server that runs massive language models without touching the cloud. Tandem LLM handles the heavy lifting on an N V D G X Spark, serving a twenty-seven billion parameter model with custom math kernels that make it run three times faster than standard setups. The real magic is in how it generates text: a smaller guessing model drafts the next words, and the big model checks them all in one pass, so you get high quality output without waiting for the slow part. It even remembers your long conversations, so restarting the computer doesn’t clear your brain.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ tasteskill/tasteskill

Taste Skill: Stop the AI Sludge

Taste Skill forces AI coding agents to build websites that actually look good, fixing the generic, repetitive designs most bots produce by default. It works by handing your AI assistant a specific set of design rules that prioritize strong layout, proper spacing, and readable text over quick, lazy shortcuts. The real magic happens when you let the AI create its own design sketches first, analyze them carefully, and only then write the code to match that vision. This simple shift stops the machine from guessing and makes it act like a real art director.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ xiayu1987/noobot

Noobot: Your Self-Hosted AI Agent with 3D Animation

Noobot is the self-hosted AI workspace that lets you animate 3D characters just by talking. This open-source tool gives you a private, isolated environment to run AI agents with real-world tools like file management and web browsing. The real surprise here is the character plugin. You can import a 3D model, tell it what to do in plain English, and watch it wave, walk, or jump right in the chat. No coding required. It saves these animations as works you can tweak later. If you want your AI to show, not just say, Noobot is a cool way to see that happen.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ tanakh/plamo-translate-desktop

Translate Anything Offline with PLaMo Translate Desktop

Stop sending your private chats to the cloud just to get a translation. PLaMo Translate Desktop lets you run a powerful AI model right on your own computer, so your words never leave your machine. Once you download the model, you can translate between fifty-three languages completely offline, meaning no Wi-Fi, no subscription, and no data leaks. It handles everything from quick chat messages to entire books, keeping the tone consistent as it goes. You can even plug it into your browser to translate whole websites in place. It is a privacy-first way to communicate across languages without worrying about who is reading over your shoulder.

πŸ†” @hackernewsgithubprojects