August 5, 2025 New releases: Claude Opus 4.1, OpenAI Open Source Model, Google's Genie 3

There were three big model releases today. The first two are interesting, but the third one is crazy. The first Anthropic's Claude Opus 4.1 was released. This is basically just an across the board update on Opus 4 model on agnostic tasks, real world coding and reasoning. It's good and it shows Anthropic's commitment to just continually improving its product. Additionally, OpenAI announced that they're releasing an open source model. And this is a pretty big place since Meta has signalled that they're out of the game and Chinese open source models are getting released and really shaking up the market. Um, interesting that OpenAI is throwing their hat in the ring. It's a reasoning model called GPT-OSS and it's available in two different sizes. So one is big, it'd be for data centers and the others would be small and run locally on someone's desktop. Running this comes with the big benefit of it being more secure since all the data would be stored locally. And not having to pay for the usage if you're the smaller model is said to be on par with OpenAI's O3 mini. The larger one is said to be similar to capabilities of O4 mini. So pretty good. So for businesses that's really secure solution, you don't have the funds to create your own proprietary model, um, and you're kind of freaked out by these other models that are open source and been popping out and you like that OpenAI brand, well that could be a solution for you. Finally, the last announcement and the craziest is that Google DeepMind announced Genie 3, which is a world building model that creates these interactive, playable environments from a single prompt. So basically you can create this simulated world through a prompt. And it's not just like video game type graphics, it's like VEO3 level hyper realistic looking. Like you could write that you're walking in a field and a waterfall behind you and you just navigate and see the waterfall that you created. There isn't public access yet, so all we have to go Google has shared. So with a grain of salt. But it does seem like the physics are pretty good and that there's object permanence. They showed a shot of this person on a jet ski going through a waterway in a town and you can see like that water from the jet stream coming up of flows out and falls like water would like very realistic splatters onto the concrete and the user can navigate, look away and look back down and the water is still on the concrete. That's hard to do and it's figured it out. It's happening. I love playing Roller Coaster Tycoon and Sims growing up and now we're on this trajectory to like create and experience those worlds. Isn't something you virtual reality. It's just like 2D images that you can navigate within this 3D space. But I mean, that's where it's going right. And so then you can really create a digital world based off of our physical world. So for businesses, definitely keep your pulse on this one because this is new immersive media.