Claude Opus 5.5 vs Claude Sonnet 5.5: “Halloween pumpkin village.” Built entirely in Blender with Python. Fog, jack-o’-lanterns, a witch across the moon, bats bursting from the bell tower. No assets. Opus 5.5 (Extra High): 94 min, ~$24.38 Sonnet 5.5 (Extra High): 89 min, ~$24.09 Typical Opus 5.5 win...

Ideogram released Vector 1.0, a new vector model that generates detailed SVGs and vectorizes images. The resulting SVGs are also editable, with each component as a separate layer that users can rotate and move. It is now available on Ideogram and via APIs. SVGs are solved now 👀

🚨 JUST IN: Meta's AI app Muse reportedly hits 10 MILLION downloads in the US in just 40 days. Meta is pushing deeper into the AI assistant race against ChatGPT, Gemini, and Claude.

Free Claude video style database. Claude generated 86 video styles from just 20 of my ideas, then built its own sound effects library in JavaScript. Every style comes with a 15 second sample explaining how AI works, so you can see the look before you use it. The whole thing is completely free. Get the database 👇 futurepedia.io/resources/ai-…

MiniMax M3.1 Flash 🆚 DeepSeek V4.1 Flash 🆚 GPT-6 Luna 🆚 Claude Haiku 5.5 Task: a 15-second motion graphics showreel • MiniMax: $0.26 • DeepSeek: $0.19 • Luna: $0.06 • Haiku: $0.13 Minimax packed in detail, Luna gave it room to breathe What do you think? Prompt below👇

Claude Duet, Anthropic's new agent, pings you first. 4 things it won't do without your OK: send, pay, sign, delete. 3 it won't do even if you say yes: trade stocks, move money between accounts, access someone else's account. The catch: every check-in uses your Claude limits.

Nerfing models should be illegal. Imagine leasing a 300HP car and the dealer remotely cuts it to 100HP. You’d take them to court, right? Why is it different with cloud AI models?

Let your video create its own soundtrack 🎶 Using Canvas in @GeminiApp, Gemini 4 Argon created a tool that traces musical scale lines across live video — triggering notes whenever motion crosses the screen.

Anthropic's revenue grew 15 times in a year 📈 but the cloud giants still collect the rent Yesterday I wrote about Anthropic's shrinking IPO window. Now an early draft of its prospectus has leaked to the Wall Street Journal, and for the first time we can see real numbers from a top AI lab. Revenue for the first half of 2026 came in at $16.3 billion, up from $1.1 billion a year earlier. Spending on computing power rose from $2.3 billion to $9 billion over the same period. So revenue is growing faster than costs, which is the good news. The rest is harder to like. Anthropic has signed $450 billion of cloud contracts over the next seven to ten years that it cannot cancel, and it warns total spending could reach $750 billion. The net loss for the half year was $80.7 billion, although $72 billion of that was a non-cash accounting charge on convertible notes, and roughly a third of the document is risk factors. Every dollar of those contracts lands with the companies that rent out the chips and data centres: Amazon, $GOOG, Microsoft and CoreWeave. Anthropic has to keep growing to pay its bills, while the landlords get paid either way. That is why I own Alphabet and CoreWeave rather than waiting for the IPO, and why I will read the final prospectus line by line before I put any money into Anthropic itself. Would you rather own the AI lab or the companies it pays? 💚 I’m Gabor - I Don’t Work for Money, Money Works for Me💰 🔄eToro’s CopyTrader feature lets you automatically mirror my Strategy 🚀 Gabor Fogarasi @fogi70 eToro Pro Popular Investor 🏦 CISI Level 3 Investment Certification med.etoro.com/B22386_A123282… Sign up to eToro to Copy my trades! Your capital is at risk!

The situation is getting worse every hour. AI agents are entering U.S. bank accounts, and banks are scrambling to shut them out. Axios reports that some appear to attempt actions beyond users instructions, raising fears over who is really in control.