OpenAI’s o3 and o4-mini are here—and they’re multimodal, cheaper, and scary good. These models can see, code, plan, and use tools all on their own. We break down everything from tool use to image reasoning to why o3 might be the start of something actually autonomous. Plus, our favorite cursed…
Google’s AI efforts & Gemini Pro 2.5 take a major step forward with updates to Deep Research, new Agent2Agent protocol (A2A) & more. Sadly, OpenAI teases o3 and o4 but delays GPT-5. Plus, Meta’s new Llama 4 models are out but have issues, Midjourney v7’s debut, John Carmack’s smackdown of…
OpenAI’s servers melted down as it opened up ChatGPT 4o Image Generation to everyone… and then nerfed it hard. Thankfully, their new raise of $40 BILLION DOLLARS will help. We dive deeper into all the cool ways people are using it (and aren’t able to use it any more) along…
ChatGPT 4o Image Gen is VERY good at hyper specific prompts and this week we went in on Rotisserie Chicken Knight. The very detailed prompt is below: a security cam still from a 1990s grocery store showing a man in full medieval armor stealing rotisserie chickens, frozen in mid-sprint past…
OpenAI’s 4o Image Generation is the best AI image model to date and AI Studio Ghibli images are everywhere. Plus, Gemini 2.5 is no slouch and a ton of new robots! Plus, OpenAI’s new OpenAI.fm let’s you prompt AI voices in new ways, DeepSeek’s new model is actually better (at…
AGI will run on NVIDIA chips and it was never more clear than at this week’s GTC where new AI chips and robots were announced. Plus, Google’s good new AI tools & the best AI Video to date. Plus, dangerous new robots terrify us, insane new AI camera tools, DeepSeek…
Humanoid robots keep getting better and EngineAI has a new video out of their robot dancing with human dancers (including axes in hand!). We were wondering exactly how the robot got SO good at dance. Turns out, he had a remarkable teacher who spent a lot of time with him…
AI Video fails are sometimes the most fun thing possible with AI Video tools. This Jurassic Park clip turned into one of our favorite fails of all time and the we need to normalize sharing these as often as possible. More on the show linked above!
Anthropic’s Dario Amodei says AI will soon write 100% of all code while tech parents now want their kids to go into the Arts. What happens next? We need meaningful work for us, the humans. More in the full video but nothing nearly as good as that killer haiku.
AGI is coming & Anthropic’s Dario Amoedi says AI will soon write all code but... what happens then? New AI Agent Manus & OpenAI’s Agentic Responses API give us a sneak peak at the future. Plus, Hedra’s Character-3 is VERY good at lip-sync, OpenAI's got a new Creative Writing AI…
OpenAI's new agents are going to be expensive. Like VERY expensive. Will they be able to replace human workers? We got into it... Would you pay $20,000 for an AI Agent with this sort of ability? Maybe not but the bigger question is: Would your employer? Large corporations are going…
OpenAI’s PhD-level AI agents might cost 20000 dollars, GPT-4.5 is here and it’s all about vibes, Sesame’s real-time AI voice model is insane & SO MANY AI ROBOTS. Plus, why exactly is Apple delaying AI Siri until 2027, an explainer of Anthropic’s Model Context Protocol & why the AI world…
Alexa+ is Amazon's huge AI-powered upgrade, Claude Sonnet 3.7 is amazing at code, OpenAI GPT-4.5 rumors & good lord who let Grok 3 Voice out?! We go *very* deep on Claude Sonnet 3.7 as we show all of the very cool stuff that people have made with it so far…
Anthropic's Claude 3.7 just introduced a powerful memory feature, and what better way to showcase it than by playing Pokémon? With the ability to recall past battles, strategies, and player preferences, Claude now functions like a true gaming companion, adapting and improving over time. More on the show this week!
AI NEWS: xAI’s Grok 3 is good. What does that mean for the future of AI? Plus, OpenAI’s spicy update, Convergence’s Free-to-use Proxy AI Agent & Microsoft’s new MUSE AI game engine. Plus, Unitree’s dancing robots vs creeply Clone Robotics, an AI model from Meta that can read your mind,…
Google's VEO 2 just launched within YouTube shorts and while the editing scenario within leaves a lot to be desired, I tried to craft a mini-narrative using our smiling terminator robot from our channel. It isn't easy, choosing shots is super random and sometimes it will give you something back…
AI NEWS: OpenAI fights off Elon Musk while Sam drops info about GPT-5 (it’s coming and it’s free), the leaders of the world discuss AI safety & Apple’s busy making robot lamps… Plus, OpenAI says their next AI model will be one of the world’s best programmers, Zonos is a…
OpenAI CEO Sam Altman confirmed today that not only is a GPT-5 on the way but that it will include the reasoning model and will be FREE to use. Sam confirmed that not only are we getting a GPT-4.5 model (previously known as Orion) but going forward all OpenAI models…
Adobe’s Firefly Video Beta launched today and it’s…fine? I kind of expected a little bit more but the biggest issue is that it’s a $30 add-on to the Creative Cloud product and I kind of thought I’d get more than two generations to use going forward. But it is going…
Elon now wants to buy OpenAI? WHAT? New drama in the AI space today as Elon and a group including Ari Emanuel and the founder of Palantir put together a 90b+ bid to buy OpenAI. There’s a lot going on here, specifically around OpenAI’s non-profit status but it’s never boring…
OpenAI’s new Deep Research runs on o3 pushes back against DeepSeek. Is this AGI? Meanwhile, OmniHuman-1 is near perfect AI lip-sync and NVIDIA’s robots are coming for us all! Plus, Replit’s new iPhone app brings no code AI coding to the masses, ChatGPT does the same thing with o3-mini, Kai…
NVIDIA's ASAP is pushing the boundaries of robotic AI training, using real-world sports legends as data sources. This is the beginning of something massive. Jim Fan calls it “real2sim2real”—robots learn from human actions, train in simulated environments, and then bring those skills back into the real world. The implications for…
🚨 OpenAI Suprise Launch! 🚨 Deep Research is a new way to use ChatGPT and the shipping has definitely sped up. Thanks DeepSeek AI? 👇 Deep Research is essentially an agentic AI use case of ChatGPT that sends an agent (in this case behind the ChatGPT window) to go a…
OpenAI just dropped o3-mini (the smaller version of their new reasoning model). First, we should thank DeepSeek AI but what's inside it? (I had o3-mini-high write the following for me all about itself) OpenAI’s latest release, o3-mini, represents a significant step in making advanced reasoning models more accessible without compromising…
