Unlike other hotly talked about AI models out there (looking at you Sora!), Luma Labs has announced that their new AI model is out and available to try today.

Available to test on the company’s website, Luma Labs reports that Dream Machine is a new AI model that is designed to make high-quality, realistic videos from text and images.

But don’t just take our word for it, you can actually go and check this new AI out yourself and make your own opinions about how it looks and how you might (if at all) potentially use it.

Introducing Luma Dream Machine

Based in Palo Alto, Luma states that their mission “is to build multimodal AI to expand human imagination and capabilities.” And this new Dream Machine AI does look like it’s here to do just that.

Designed to be a highly scalable and efficient transformer model trained directly on videos, this new AI should be capable of generating physically accurate and consistent shots of pretty much whatever your heart desires.

Create Action-Packed Shots and Camera Moves

A core selling point of this new Dream Machine AI, at least from these early demos, seems to be that this is a very cinematically-minded model that understands how people, animals, and objects interact with the physical world—which allows for users to create videos with character consistency and accurate physics.

At launch here Dream Machine offers the ability to generate 5-second shots with realistic smooth motion, cinematography, and all the drama you can muster out of it. It’s also incredibly fast compared to other models and can generate 120 frames in 120 seconds.

It also looks like Dream Machine can help you experiment with what might be an endless array of fluid, cinematic, and naturalistic camera motions matching the “drama” of the scene you give it to create.

Try it Yourself

Of course, by far the most noteworthy point here is that Dream Work is actually live and available for anyone to test today. Which you can do here. So, if you’re interested, give it a shot. Or don’t—it’s up to you.