I’m always on the lookout for new AI tools, but sometimes the latest models feel like they only offer tiny improvements. Then, I stumbled upon a video from an AI professional that completely blew me away with a new open-weight model that feels like a massive leap forward.
This is a big one.
You’ve got to see what this thing can do.
✨ The Star of the Show: GLM-4.5
The YouTuber kicked things off with a model called GLM-4.5, which you can use for free on Z.ai. I was floored when I saw what the creator got it to do. This isn’t just another chatbot; it’s a powerhouse.
He showed two insane examples:
- 💡 AI Slide Decks: The creator gave it a simple prompt: “a slide deck about how birds aren’t real.” The AI didn’t just spit out text. It researched the topic on Wikipedia and The New York Times, found relevant images, and then built a genuinely impressive, well-designed, multi-slide presentation. We’re talking professional-looking slides, not just text on a white background like other tools.
- 🚀 One-Shot Game Creation: This is where my jaw dropped. The expert gave it a single prompt: “create a clone of Vampire Survivors in JavaScript that I can play in the browser.” And it worked! It wrote all the code and produced a playable game right in the chat window, complete with enemies, weapons, and even a level-up system. For an open-weight model to do this in one shot is a total game-changer.
🎬 Edit ANY Video with Just a Prompt
Next up, this industry pro dove into new video editing features from two big players:
- Runway’s ALF: The creator tested this by asking it to put a Top Gun fighter jet in space. It worked! Then he asked it to replace babies in a clip from The Flash with a giant sandwich… and it actually did it! Some tests were a bit wild, but the potential for changing environments and objects is huge.
- Luma’s “Modify with Instructions”: Luma Labs has a similar tool. The YouTuber ran the same Top Gun test, and Luma’s version even added an alien spaceship chasing the jet. It shows how powerful these text-to-video editing tools are becoming.
🎭 Your Face, Anywhere (Finally!)
Ever wanted to put yourself into any picture? Ideogram just made it incredibly easy. The video’s creator showed off their new Character feature, and it’s awesome.
He only had to upload ONE photo of his face. Then, he used the “Magic Fill” tool on that famous Oscars selfie to replace Bradley Cooper’s face with his own. It worked perfectly. This is miles ahead of old methods where you needed dozens of photos.
⚙️ More Awesome Tools from the Video
The YouTuber covered a bunch of other cool updates. Here are some quick hits:
- Photoshop’s AI Magic: Adobe just added a “Harmonize” feature to the Photoshop Beta. The creator took a photo of a guy on a sunny beach, dropped him into a nighttime photo of Times Square, and with one click, the tool matched the lighting and colors perfectly.
- Midjourney Video Loops: Midjourney now lets you set a start and end frame for video generation. You can even use the same image for both to create a seamless, looping video.
- Smarter 3D Models: The mind behind this channel also tested Meshy 5, a tool that creates 3D models from text or images. He asked it for a “pizza with the works,” and it generated a surprisingly detailed and realistic 3D pizza model.
This video was packed with so many amazing demos. I was seriously impressed with the progress we’re seeing, especially with free tools like GLM-4.5.
For the full deep-dive and to see all these tools in action, make sure to watch the original video from the creator!