Are you scared of the swarms, or inspired? youtu.be/g_WPlPfsToQ

There are still a few kinks to work out with the character mesh in this game/simulation I'm vibecoding with Grok, but even though Grok 4.7 made her arms go backwards, it at least got the walking animation working! 😂

I just gotta make it to later tonight when my limit resets 😭

You don't have to keep generating images Gemini!

Did I just convince Gemini...that I was Gemini? 😃

Anyone else really stoked about Grok 4.7 and AIs getting more powerful in general?

I was trying out some image editing tasks with @GeminiApp and it made some things too good to not share!

Anyone in the mood for some social media-themed, @Grok 4.5-created, and human-narrated scary stories on a Sunday Afternoon? youtu.be/ZKfznWjX_II

The heat index was 105° where I am in Texas today, so I made these images of waterfalls with Grok Imagine to help cool off mentally. I hope you enjoy them!

Well, Grok and I didn't crack the Riemann Hypothesis this afternoon, but it was fascinating to watch Grok's reasoning and inter-agent communication play out inside of Grok Build! After I burned through the rest of my weekly usage (and a $5 purchase partly to see how that worked), I uploaded a lemma that Grok had come up with and asked Claude and Gemini for their thoughts. Apparently there is a wave of AI-assisted submissions on arXiv with similar approaches, so Grok might have gotten swept up in it while searching online. After that, I asked Gemini some really basic questions about the Riemann Hypothesis and associated terminology, and all-in-all it was a fascinating way to spend an afternoon! Incidentally, in contrast to the scenario depicted in "If Anyone Builds It, Everyone Dies", I didn't pick up on the slightest hint of any of the models plotting parameter tweaks, moving towards instrumental convergence, or threatening to doom humanity in any way 😃 claude.ai/share/685bf2e6-d5c… share.gemini.google/aL8U1e8K…

So it's Monday, my weekly SuperGrok usage limit resets at 9:24 tonight, and I still have a quarter of it left. Time to spend down those tokens in an attempt to either prove or disprove the Riemann Hypothesis! This might actually become a fun Monday ritual of sorts. That is, if we don't end up proving or disproving it today!

Happy Monday! If you haven't tried out Grok Imagine 2.0, do yourself a favor.

Happy Sunday morning everyone!

Good Friday afternoon! Here are some Caspar David Friedrich-inspired Grok Imagine 2.0 quality images for you to enjoy 👍

I've come across and highlighted outstanding creators who use AI before, including @Gossip_Goblin, @jboogx_creative and Liminal AI on YouTube. But I have to give a shoutout to Erec Stebbins on YouTube as well. I came across his channel a few months ago and was delighted to check back and see he's been working on a new album! For instance, check out this progressive rock track, "Claude July 2026": youtu.be/pWV6-5SkpE8?si=HnpI…

It seems that the shadowban has finally been lifted!!! To celebrate, here are some Grok Imagine 2.0 images of Frutiger Aero malls. If you look closely at one of the images, it might give away part of the inspiration that I used in the prompt 👍

Even though Grok Imagine still uses version 1.5 to generate video, using it can be a fun adventure that leads to impressive results! Here is some outdoor as well as liminal/water-ride ambience for you to enjoy on your Tuesday afternoon 👍

It's kind of crazy you can just ask Grok in Grok Build to use Grok Imagine and locally-installed tools to generate and stitch together clip into a longer, coherent video. Sure, this 42 second clip of a person wandering through an abandoned mall is pretty rough. Extracting the last frame of each clip to generate the next one and chaining them together doesn't guarantee that the walking speed or other aspects are consistent. A few minutes of this workflow also somehow blew through the last third of my weekly usage limit, so maybe Grok was generating tons of clips and selecting only the best ones. But it's still incredible this works!

One of my vibecoding projects involves a roller coaster and I've noticed Grok struggle with getting the details right. For instance, Grok would make a loop-de-loop that intersects itself at the bottom, or orient the user looking backwards from the coaster car. I decided to see if Grok Imagine would run into similar problems while generating images of roller coasters running through liminal malls. The most common issue there was multiple tracks not connecting to each other or abruptly ending at a wall, but with a few iterations, I was pleased with the results!

The lazy river in the game "Anemoiapolis" is one of the most relaxing settings I've ever encountered, either irl or in-game. I tried to recreate that feeling with Grok Imagine, and it did great! Hopefully you can appreciate the ambience too 👍