At Luma AI, we're building multimodal general intelligence, starting with the world's best video models, and we've raised $200 million from investors like Andreessen Horowitz and Nvidia.

AI models are increasingly showing multimodal capabilities, suggesting they might develop more complex experiences in the future.

The convergence of reasoning models and multimodal models will lead to a new kind of computer interface that can reason, generate real-time video, and provide personalized experiences.

By training models that learn like the human brain, we aim to generate entire worlds and overcome the limitations of large language models.

Models will get better at understanding and creating images, audio, and especially video, which will dramatically enhance how we interact with data.

To create characters like Morales, we developed a unique virtual production tool that combines live streaming video and audio with motion capture technology, allowing us to build a storytelling studio.

We aim to provide a platform that allows users to build on our models, offering tools for customization and control to create effective products.

Gemini was built to be natively multimodal, meaning it can take in images, audio, video, and more, which is essential for progressing towards Artificial General Intelligence (AGI).

To build intelligence that collaborates with humans, it must be trained on diverse data, including video, as human learning occurs across multiple modalities.