Trying to keep up with all the AI news can feel like drinking from a firehose. Well, this week, that firehose basically exploded. I was trying to make sense of it all when I stumbled upon this incredible video from an AI professional that recaps what might be the biggest week for AI in 2025.
This innovator breaks down a dizzying number of game-changing releases, and I had to share the highlights with you.
🚀 OpenAI Drops a Double-Bombshell
OpenAI was the star of the show this week with two massive releases.
- GPT-5 is Here: This new model is a beast. The video’s creator explains that it has “PhD-level intelligence” and crushes all the benchmarks, especially in coding. The interface is simpler, it’s faster, and it’s rolling out to everyone, including free users.
The YouTuber even showed how the model built a functional Vampire Survivors clone game from a single prompt! If you don’t have it in your ChatGPT account yet, the expert points out that it’s already live inside Microsoft Copilot. - GPT-OSS (Open Source): Just before dropping GPT-5, OpenAI released two powerful open-weight models. This is huge because you can download them, run them locally on your own machine (even offline!), and your data stays completely private. They’re about as powerful as the top models were just before GPT-5 came out.
✨ Anthropic vs. OpenAI: The Coding Showdown
For about 24 hours, Anthropic’s new Claude Opus 4.1 was the best coding model on the planet. This industry pro put it to the test, asking it to build a fitness app. Claude generated a super impressive app with multiple routines, timers, and even audio feedback.
Then, the expert gave the exact same prompt to GPT-5. While GPT-5’s app was simpler and less feature-rich, it highlights how we now have multiple top-tier models to choose from. If one gets stuck, just hop over to the other!
🌍 Google’s Wild New Worlds
Google DeepMind unveiled Genie 3, and it’s honestly mind-blowing. It’s a “world model” that can generate interactive, playable video game levels from a single text prompt or image.
The person who shared it showed demos of users navigating custom-made worlds: flying helicopters, exploring planets, and walking through fantastical scenes, all generated in real-time. Just wild.
Google also rolled out a Gemini Storybook feature. The creator tested it by making a hilarious, slightly-NSFW storybook for adults about a dad tired of cleaning up toys. It worked perfectly, with consistent art and story pages.
⚙️ The Rapid Fire Round-Up
As if that wasn’t enough, here are a few other things this channel covered:
- Nvidia AI Twin: A new feature that lets you upload a video of yourself to create a realistic digital clone for use in AI-generated videos.
- Grok Imagine: Now creates images and videos with audio. Elon Musk even made it free for all US users for a few days (the video notes a VPN might work for others 😉).
- ElevenLabs Music: You can now generate full songs with vocals and instruments right inside ElevenLabs. The YouTuber showed how you can even create instrumental backing tracks to jam along with.
- Leonardo AI Updates: The platform now has a stunning new image model called Lucid Origin and has integrated Veo 3 for image-to-video generation.
I was blown away by how much happened in just a few days. This is just a quick summary of a jam-packed video.
For the full deep-dive and to see all these demos in action, make sure to watch the original video from the creator. It’s the best roundup you’ll find!