Programming eBooks

360 AI Chapters online and 3600 quizzes at the end of chapters: C# & AI: programmingcentral.vercel.app/books/csharp/ Typescript & AI: programmingcentral.vercel.app/books/typescri… Choose Volume/Chapter on the left menu and enjoy ! Thanks !

@mervenoyann @huggingface Wow 1450

future is local 🔥 Google DeepMind just released Gemma 4: local frontier in many sizes, all modalities with free license 🤯 we ship Gemma 4 in transformers, llama.cpp, transformers.js and more for your convenience 🫡 plug-and-play with your agents 🙌🏻 read our blog ⤵️

New ebook. Apple SwiftUI for AI Apps. Building reactive, intelligent interfaces that respond to model outputs, stream tokens, and visualize AI predictions in real time leanpub.com/SwiftUIforAIAp…

New ebook. Apple Swift: Natural Language & Speech. NLP, sentiment analysis, text classification, and Speech-to-Text with Apple's Natural Language and Speech frameworks. leanpub.com/NaturalLanguag…

@cheikhshift True

@ProgrammingCCC That's actually impressive - Chinese AI labs are really stepping up their game lately!

Openrouter.ai most popular models all from China !!

New ebook. Rust programming. Advanced Memory Patterns for AI. Mastering Lifetimes, Smart Pointers, and custom allocators for managing large models and datasets leanpub.com/RustAdvancedMe… amazon.com/dp/B0GRVLKP76

New ebook available: Apple Intelligence & Foundation Models. Building apps with Apple's on-device LLM APIs, Writing Tools, and the Apple Intelligence framework: leanpub.com/AppleIntellige…

@XFreeze Grok 4.20 hitting #1 with 78% is a massive flex. It’s officially the end of the 'hallucination era.' Seeing GPT-5.4 (xhigh) and Claude 4.6 getting gapped like this shows xAI’s 'Truth-First' approach is scaling way better than the competition. The leaderboard just got a new king.

Most AI models hallucinate more than you'd think and make up stuff that doesn't exist Grok 4.20 just ranked #1 in Non-Hallucination Rate with a 78% score - beating Claude Opus 4.6(max), Gemini 3.1, GPT-5.4(xhigh), and every other model on the list xAI is quietly winning the accuracy game… and it’s built to be truthful

@thisguyknowsai This is the core of the JEPA vs. Generative debate. While everyone is obsessed with 'pixel-perfect' generation, Meta is proving that internal representation is where the real 'intelligence' lives. What happens when we scale this to 20 million hours?

🚨 BREAKING: Meta researchers showed a model 2 million hours of video. No labels. No physics textbook. No supervision at all. It learned gravity. Object permanence. Inertia. And it just beat Gemini 1.5 Pro and GPT-4 level models at physics understanding. Here's what just happened:

@seikixtc Wait, terminal shows 202B params while you mentioned 1T—is this a specific MoE config? Either way, seeing that 'Streaming active' log on an M2 Max is insane. What kind of tokens/sec are you hitting once the budget is exceeded? 🤯

I got a 1T-parameter model running locally on my MacBook Pro. LLM: Kimi K2.5 1,026,408,232,448 params (~1.026T) Hardware: M2 Max MacBook Pro (2023) w/ 96GB unified memory Running on MLX with a flash-style SSD streaming path + local patching. This is an experimental setup and I haven’t optimized speed yet, but it’s stable enough that I’ve started testing it in an autoresearch-style loop. #LocalAI #MLX #MoE

@julien_c @huggingface Superb

hf-mount Attach any Storage Bucket, model or dataset from @huggingface as a local filesystem This is a game changer, as it allows you to attach remote storage that is 100x bigger than your local machine's disk. This is also perfect for Agentic storage!! Read-write for Storage Buckets, read-only for models and datasets. Here's an example with FineWeb-edu (a 5TB slice of the Web): 1️⃣> hf-mount start repo datasets/HuggingFaceFW/fineweb-edu /tmp/fineweb It takes a few seconds to mount, and then: 2️⃣> du -h -d1 /tmp/fineweb 4.1T ./data 1.2T ./sample 5.3T . 🤯😮 Two backends are available: NFS (recommended) and FUSE Let's f**ing go 💪

@NewsFromGoogle @FastCompany Wow

Google was just named #1 in the @FastCompany 2026 World’s Most Innovative Companies list. 🎉 Google is also ranked #1 in their Artificial Intelligence category. See the full story. fastcompany.com/most-innovativ…

New ebook avaialble: Apple Core ML & Vision Framework. On-device image classification, object detection, and custom model integration with Core ML and Vision. leanpub.com/CoreMLVisionAI… New: 18 ebooks Online programmingcentral.vercel.app click on "Read chapters" buttons. 360 chapters

@scaling01 Try Xiaomi mimo v2 ...

GPT-5.4-mini is dead on arrival you can just use Kimi-K2.5 for $1.5 less and better performance

@ChristosTzamos Some implications programmingcentral.vercel.app/blog/posts/the…

1/4 LLMs solve research grade math problems but struggle with basic calculations. We bridge this gap by turning them to computers. We built a computer INSIDE a transformer that can run programs for millions of steps in seconds solving even the hardest Sudokus with 100% accuracy

@twentyvisionai @joemccann And better security propably

This is the most underrated direction in AI right now. Instead of making LLMs better at arithmetic through more training data, just embed a deterministic computation engine in the weights. The implications go way beyond Sudoku — imagine financial modeling, physics simulations, or cryptographic operations running inside inference. Hybrid architectures where neural nets handle reasoning and embedded interpreters handle precision.

This is actually insane. Dude hard-coded a WebAssembly (WASM) interpreter into the weights of a transformer, losslessly. In essence, a computer is running inside a LLM that can actually run computations, not infer or guess a calculation like most do today.

@joemccann Intuitively it seems like something worthy of a Turing Award to me.

@ChristosTzamos Nobel prize

4/4 Read more at our blog post: percepta.ai/blog/can-llms-…