Theo's blog

Free AI from NVIDIA: a practical guide to build.nvidia.com

NVIDIA lets you try dozens of top open models for free, including Kimi K3, DeepSeek V4 Pro and Nemotron, plus agent skills, blueprints and self-hostable NIM containers. Here's what's free, how to start in five minutes, and the catches to know.

Nominated as Business Innovator: a letter I opened standing up

The German Innovation Institute, together with UNTERNEHMER magazine, has nominated my company as a Business Innovator for askTheodor. A nomination, not an award yet, but after years of building it alone, it means a lot. Thank you for being part of it.

Mistral Large 4: Europe finally has a frontier model of its own

Mistral Large 4, “le Chonk”: 1 trillion parameters, trained and served from Mistral's own European datacenters, open weights by the end of October. Why a sovereign, GDPR-friendly European model matters for every business that handles personal data.

Nvidia's AI Empire: How Jensen Huang Built the Future of Computing

Kimi K2 Thinking: Open‑Source Reasoning Hits the Frontier

Moonshot AI’s Kimi K2 Thinking pushes open‑source reasoning to frontier‑level benchmarks with tool-augmented performance, fast INT4 inference, and real-world agentic coding gains.

Grokipedia went Live yesterday

xAI launches Grokipedia (beta v0.1), a Grok-powered encyclopedia with 885k+ articles and real-time fact-checking — raising big questions about neutrality vs. speed.

Gemini Enterprise: The New AI Business Experience

Google unveils Gemini Enterprise at their Gemini at Work event, promising to revolutionize workplace AI beyond simple chatbots with a comprehensive platform that integrates company data, tools, and people.

Google Limits Search to 10 Results Per Page: SEO, AI, and Visibility

Google quietly removed support for showing 100 results per page. The change is reshaping SEO data, AI training inputs, and who gets discovered online.

Meet the new Ray-Ban Meta Glasses

Discover the revolutionary Ray-Ban Meta Display Glasses with AI capabilities, EMG wristband control, and breakthrough AR technology that's changing how we interact with the digital world.

AI Video Generators - Unleash your Creative Powers

Explore the top AI video generators that can help you create stunning videos from text, images, and more, unleashing your creative potential.

🚀 Launching My Virtual Try-On WooCommerce Plugin: Revolutionary AI Technology That Works With Regular Photos

After months of development, I'm excited to announce the launch of my Virtual Try-On WooCommerce plugin - a game-changing solution that brings AI-powered virtual try-on to any e-commerce store using regular HD photos, no expensive professional photography required!

Qwen: Powerful AI Models for Multimodal Tasks

Exploring Qwen's capabilities in language, vision, and audio processing.

Top AI Image Generators in 2025: Strengths, Weaknesses, and Pricing

A concise, practical comparison of today’s leading AI image generators—what each does best, where they struggle, and how they’re priced.

BREAKING: Google's Nano-Banana - The FREE AI Image Editor Outperforming Photoshop

Google's Gemini 2.5 Flash Image model, dubbed 'Nano-Banana', has anonymously emerged as the world's #1 AI image editor, offering Photoshop-level capabilities in seconds, for free.

Custom AI Chatbots: Revolutionizing Global eCommerce Customer Service

Discover how custom AI chatbots are transforming eCommerce with multilingual support, 24/7 availability, and personalized customer experiences.

DeepSeek V3.1: Revolutionary Hybrid AI Model Cuts Costs by 6x While Outperforming GPT-5

Chinese startup DeepSeek AI just released V3.1 - a groundbreaking hybrid model with dual Think/Non-Think modes that's 6x cheaper than GPT-5 while delivering superior performance.

Dual Web: Optimize Your Site for AI and Humans

Discover how Dual Web helps brands serve optimized content to both human visitors and AI systems for better visibility and accuracy.

Twos AI: Supercharge Your Productivity with an Intelligent Personal Assistant

Deep dive on how Twos uses AI PALs to consolidate notes, reminders, and to-dos for a more organized workflow.

MovieFlo.AI: Revolutionizing Cinematic Campaigns with AI Magic

Spotlight on how MovieFlo.AI can turn ideas into campaign-ready cinematic content, featuring insights from Lucasfilm and ILM vets.

🐳 12 Years of Docker: Shipping Projects Anywhere

Reflecting on over a decade of using Docker containers to build, ship, and run projects seamlessly across environments. Why Docker remains my favorite tool for development, deployment, and AI workflows.

Container Server Nodes in Orbit: The Next Revolutionary Step?

My thoughts on a crazy idea that might change everything: 2,800 satellite server nodes as the first step in a new global computing market from space.

The humanoid Robot Revolution is Real and it begins now.

From factory floors to family rooms, humanoid robots are crossing the threshold—driven by home‑first design, safe tendon-driven hardware, and learning loops that feed AGI ambitions.

The first time the AI won the humans and a championship.

In 1997, IBM's Deep Blue defeated Garry Kasparov in chess—the first time an AI beat the reigning world champion in match play. Here's a comprehensive timeline of AI's most important milestones from that historic moment to 2025, including George Delaportas's pioneering GANN framework from 2006.

Vibe-Coded Websites and Their Technical Weaknesses

AI-generated websites look stunning but often ship with basic technical issues that hurt their performance and accessibility. Here's what I discovered after analyzing 100 vibe-coded sites.

Why Students Must Focus on Physics and Math (Not Just Coding) in the Age of AI

In today’s rush to learn coding and prompt engineering, a deeper truth is getting lost — the future of AI belongs not just to coders, but to thinkers who understand the underlying laws of nature and logic.

Jules: My New Favorite AI Coding Assistant

A personal review of Google's Jules, the asynchronous AI coding agent that has become an indispensable part of my workflow. It's free, powerful, and surprisingly reliable.

OpenAI re-activates their old models of chatGPT after Sam Altman admitted a problematic GPT-5 launch

In a surprising turn of events, OpenAI has re-enabled access to legacy ChatGPT models following widespread user complaints about GPT-5's inconsistent performance and decision-making flaws.

GPT4All: Your Friendly Local AI App (Free & Open Source)

A casual, beginner-friendly guide to running AI models locally with GPT4All — no fancy hardware required.

🚀 GPT-5 is Here: Smarter, Faster, Safer

OpenAI has officially launched GPT-5, the most advanced iteration of its language model. Here's everything you need to know about the new capabilities, features, and improvements.

Free Local AI Development with Bolt.AI and Ollama: Code Without the Cloud Costs

Learn how to set up and use Bolt.AI with Ollama to run powerful AI coding assistance completely offline on your local machine, without any subscription fees.

Developing a Custom AI Assistant for E-Commerce

Learn how custom AI chat assistants can transform your e-commerce business with 24/7 customer support, personalized shopping experiences, and increased sales conversions.