Today we’ve open-sourced ‘Orb’, the state of the art AI model for simulating advanced materials. Built upon our proprietary foundation model (LINUS), Orb outcompetes models from Google and Microsoft on accuracy and speed.
When we started Orbital, our goal was to leverage AI to accelerate the creativity and ingenuity of scientists creating the advanced materials that will power the energy transition. One way to do this is to use computers to peek into the inner functioning of materials - to simulate (“in silico”) at an atomic level what would be impossible to view under a microscope. By understanding the mechanisms that give advanced materials their extraordinary properties, we can use computers to design more performant materials.
But, and this should come as no surprise, simulating quantum physics is hard! Traditional approaches have been limited by the huge computational resources required, with the consequence being that you can only simulate a drastic simplification of what is really going on at an atomic level.
AI provides a new approach. By substituting slow, costly traditional simulations with rapid, AI-driven alternatives, we can achieve unprecedented fidelity in understanding atomic dynamics - and, crucially, figure out how to change a material (whether a catalyst, battery, or semiconductor) to improve its performance
Today we are proud to release “Orb” - the world’s best AI model for advanced material simulation. It is more accurate than comparable models from Google and Microsoft and 5x faster for large-scale simulations than the leading available alternative. We are releasing Orb under a permissive open-source license - free for non-commercial uses and startups, to maximise the impact of this technology and accelerate development efforts of teams around the globe. You can find a full technical description on our github repository.
Orb is a fine-tuned version of our internal foundation model, called ‘LINUS’, that we’ve trained from scratch. We have a technical blog covering key elements here, and a full technical report planned for the near future. Stay tuned over the coming months for more releases.
Finally, I’m incredibly proud of our team - a small, tightly-knit group with a great deal of focus. Such an achievement serves as a reminder that a scrappy, highly motivated startup can still compete, even in an age where AI is so resource-, power- and people-intensive.
If you’re interested in joining our team, visit our careers page to explore open positions. Don’t see a role posted? We would love to learn more about you and your experience. Feel free to introduce yourself and send us a CV at [email protected].