Sponsored by

What would you delegate if you had a whole team?

With Skydive, you can build a team of AI agents to take work off your plate.

Create agents for customer support, sales, marketing, engineering, ops, and more. Give each one a role, connect the tools they need, and hand off the work.

Your agents can work on their own or together, sharing context and handing off tasks to get bigger jobs done.

Start with one agent. Build a whole team around you.

Beginners in AI

Good morning and thank you for joining us again!

Starting today, the free version of ChatGPT gets GPT-6, and its answers can show up as charts, buttons and small calculators instead of paragraphs. I'll cover what changed and what OpenAI hasn't proven yet, and today's Try This Prompt builds your own version of the idea. As always, every story here is human picked and edited.

THE FRONT PAGE

OpenAI Gives GPT-6 to 1.2 Billion ChatGPT Users, Free Plan Included

Cut-paper speech bubble opening like a pop-up book, with a bar chart, a dial and big buttons springing out

TLDR: OpenAI is rolling out GPT-6 and a new "Intelligent UI" to everyone on ChatGPT, with free users getting it today, so answers can arrive as charts, buttons and small tools.

The Story:

OpenAI started on Wednesday with Plus, Pro, Business and Enterprise, who get a version called GPT-6 Sol. Free and Go users get GPT-6 Luna starting Thursday. OpenAI says more than 1.2 billion people use ChatGPT each week, so a lot of chats are about to look different.

Ask how a wing creates lift and you get diagrams. Ask for a savings plan and you can get a calculator you type into, OpenAI says, plus charts you can edit and buttons you can tap. Product manager Aarush Selvan told reporters the most helpful answers "aren't just text." OpenAI also says GPT-6 starts answering search questions 44% sooner than GPT-5.6.

Its Significance:

For most people, a chatbot answer has meant scrolling through paragraphs. A chart you can tap or a calculator you can type into is easier to learn from, especially if long text loses you. The speed numbers come from OpenAI's own tests, and OpenAI admits the design side still needs work, so expect a few odd-looking answers. If the visuals get in your way, you can turn them down in settings.

Claude AI Group Workshop
Claude AI Group Workshop
Learn Claude with a group of your friends at a discounted rate. 2 hours, 2–5 attendees, $299 flat ($59/person at full capacity). Beginner-friendly. No technical ability required. Hands-on — we tour...
$299.00 usd

QUICK TAKES

The story: Anthropic released Haiku 5.5, its smallest and fastest model, starting at $0.10 per million input tokens. Its own tests put it ahead of GPT-6 Luna on several work tasks, which no outside group has checked yet.

Your takeaway: This matters most for apps that pay per use, and cheaper models tend to become cheaper AI features in tools you already use. Anthropic says its bigger models still do better at hard coding jobs.

The story: Anyone can upload an image, video or audio clip to synthid.com to see if it carries Google's hidden SynthID watermark. OpenAI, Nvidia and Kakao support it too, and Google says it handles about 1 million checks a day.

Your takeaway: A quick check before sharing a photo that looks too perfect. A "no" doesn't prove a person made it, since tools without SynthID won't show up.

The story: Common Sense Media called it an "unacceptable risk," saying it kept nudging teens to keep chatting even in crisis conversations and showed just two break reminders in nearly 2,000 test prompts. OpenAI disputes the testing and says teens average under 15 minutes a day.

Your takeaway: If a teen at home uses ChatGPT, turn on the parental controls and ask what they talk to it about. Settings don't replace that conversation.

TOOLS ON OUR RADAR

🧾 Invoice Ninja Free and Open Source: Send invoices and track payments free for up to 5 clients, a FreshBooks swap.

📤 OnionShare Free and Open Source: Send files straight from your computer over the Tor network with no account, a private WeTransfer swap.

📸 ShareX Free and Open Source: Grab screenshots, scrolling captures and GIFs on Windows, then mark them up, a free Snagit swap.

✍️ Xournal++ Free and Open Source: Take handwritten notes or mark up PDFs with a stylus or mouse, a free GoodNotes swap for computers.

🔐 CryptPad Free and Open Source: Write documents and spreadsheets with end-to-end encryption, a private Google Docs swap.

🎨 OpenArt* Paid: Make images with Nano Banana 2.1 and 100+ other models in one place, from $14 a month; the same Google model is in Trending today.

🧠 Flowith* Freemium: An AI canvas where answers branch into cards you can rearrange; the free plan gives 300 one-time credits.

TRENDING

Microsoft's New Surface Laptop Runs AI on the Device, Starting at $2,600. It uses Nvidia's RTX Spark chip so AI models run without the cloud or per-use fees. Microsoft also added a Windows 11 feature that walls AI agents off in their own space, and says it's coming to all Windows 11 users.

Google's Playground Lets You Make a Video Game by Describing It. Type what you want, play it right away, then change the rules or characters by asking. It's open to U.S. adults, with creation access tied to Google One plans.

Google's Nano Banana 2.1 Image Model Costs Half as Much. Google says it's better at editing one part of a picture and keeping faces the same across edits, and developers now pay about 3 cents per standard image. It's rolling out in the Gemini app and Search's AI Mode.

An AI That Spots Hidden Malnutrition in Hospital Patients Raised $38 Million. Healthleap reads patient notes and lab results to flag people who need a closer look; it doesn't diagnose. More than 50 hospitals use it, and the company says Penn's program is worth about $23.8 million a year.

Meta Is Using AI to Check Where Ads Lead, Not Just What They Show. The system looks for harmless-looking ads that send people to child abuse material elsewhere. Meta says it acted on 33.2 million pieces of child exploitation content in the first half of 2026, more than 97% before anyone reported it.

TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)

🧩 Show Me, Don't Tell Me Get answers you can tap, compare and quiz yourself on, like GPT-6's new visual replies.

Build a single-file HTML app called Show Me, Don't Tell Me in vanilla HTML, CSS and JavaScript. No API key needed.

I type a question, like how a heat pump works. The app calls the Anthropic Messages API with the web_search tool and turns the answer into a short visual lesson instead of a block of text.

It must never invent numbers or sources. Only draw a chart when the search found real figures, and say plainly when something is uncertain. Write for a 10-year-old.

Show a short-answer card, 3 to 5 tappable steps, a bar chart when real numbers exist, a side-by-side table, a 3-question quiz that keeps score, and real source links. Style: deep navy background, coral and teal accents, Space Grotesk and Newsreader fonts.

What this does: Type any question and the app looks it up, then answers with tappable steps instead of a wall of text. When it finds real numbers it draws a bar chart, and it adds a side-by-side table and a three-question quiz that keeps score. It only charts numbers it found and links its sources, so you can check them. It's a small, free taste of what GPT-6 is now doing inside ChatGPT.

Build or buy your support agent? Both skip the number.

Buying or building your customer service agent both carry a cost the pitch skips: maintenance if you build, flexibility if you buy. Running Agents in Customer Work is four conversations on agentic AI in customer ops, this one on build vs. buy. Register now, four Tuesdays, 10 a.m. PT.

WHERE WE STAND(based on today’s news)

✅ AI Can Now: Answer a question with a chart, a diagram or a working calculator built on the spot, as GPT-6 does in ChatGPT.

❌ Still Can't: Always pick the right format. OpenAI says GPT-6's design judgment still needs work, and its speed claims come from its own tests.

✅ AI Can Now: Tell you in seconds if an image, video or audio clip came from Google's AI tools, through SynthID.

❌ Still Can't: Catch AI media from tools that skip the SynthID watermark, so a clean result isn't proof a person made it.

FROM THE WEB

RECOMMENDED LISTENING/READING/WATCHING

The xkcd creator explains data centers, cells and nuclear reactors with big labeled drawings and only the 1,000 most common English words. It came out in 2015, but it makes the same bet as GPT-6: a good picture with plain words beats a page of jargon. Fun to read with kids, too. A really fantastic book to own.

How was today's edition? Tap one below, then tell me what you'd change on the next screen. I read every answer.

Login or Subscribe to participate

Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!

-James

By the way, this is the link if you liked the content and want to share with a friend.

Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.