Grok 4.7 is out and it's gunning for the big models 🤖

Good morning. xAI dropped Grok 4.7, its new flagship model. A ChatGPT researcher launched Jev, an AI that supposedly "can't hallucinate," though the claim is sneakier than it sounds. And a four month old stick figure channel is pulling millions of views on basically zero production. Would you run a faceless channel if you never had to show your face or film a thing? Hit reply and let me know! Let's get into it.

xAI dropped Grok 4.7, its new flagship model, and it's live right now with no waitlist in the Grok app, the API, Cursor, and GitHub Copilot. If you use Grok, you basically already have it.

It's a real step up from the last version, bigger, better at sticking with long multi-step tasks, and it double-checks its own answers more before handing them over. It's tuned mainly for heavy knowledge work, and on the benchmarks it lands just behind the top models from Anthropic and OpenAI, which is the whole story here, xAI closing the gap on the frontier faster than most people expected.

For creators, 4.7 is the new brain in the Grok app. If you use Grok for the writing side of your work, scripting, brainstorming hooks, researching a topic, this is a genuine upgrade. Worth a spin if you haven’t tried it yet.

You might've seen headlines about Jev, a new model from ex-OpenAI researcher Diogo Almeida that supposedly can't hallucinate. It raised $40 million, and the claim is technically true. It's also not what it sounds like.

Here's the catch. Jev can't hallucinate because it doesn't write sentences at all. Instead of generating text, it just picks from a set of answers you define ahead of time, like yes/no, a category, or a score, and hands back a confidence number. So it literally can't make anything up, because it was never writing in the first place. It's built for people to get quick answers, for brainstorming content, more like a smart filter than a ChatGPT replacement.

Boxing a model into valid outputs instead of letting it freestyle is the same instinct behind where image and video AI is going, models that stay locked to your reference so a character's face or a product doesn't morph between shots.

Here's a faceless channel worth studying. Ink Explainer is four months old, has around 93,000 subscribers, and posts simple educational explainers like "What Did Ancient Humans Actually Do All Day?" The catch is there's barely any production to it. No face, no fancy animation, just stick-figure drawings on screen with a voiceover. That one video is past 7 million views.

It works because the format does the heavy lifting. The titles are pure curiosity bait, questions everyone's low-key wondered about, and the plain stick-figure art actually helps, it looks approachable and keeps you locked on the story instead of flashy visuals. The owner runs the whole thing like a pipeline with almost none of their own time in it, and has already cloned the format to a second channel.

Here's the playbook. Pick a niche full of "huh, I've always wondered that" questions, answer them in tight scripts over simple recurring drawings and a clean voiceover, and curiosity-driven titles. The visuals staying dead simple is a feature, not a shortcut, it's faster to make and easy to repeat. Once one format hits, you copy the machine into the next video.

Stuck on a tool or have an AI video question? Reply, I’ll do my best to help.

Created something cool? Send it over and I might feature you. 📺

📫 Want to get your tool featured or advertise to our readers? Let's chat → [email protected]

Thanks for tuning in!

-Kevin