Orb models represent near SoTA performance across a range of evaluations with an up to 10x reduction in latency and 8x reduction in memory for our lightest weight model. We've also trained conservative models for the first time, representing another point on this Pareto frontier.

Great to see progress by Orb in improving the speed and accuracy of their foundation machine-learning potential — and that our k_SRME metric has been helpful in the effort!

We've released our new foundation models for simulation. Our most accurate model sets a new state-of-the-art, while our fastest model is 10x more efficient — and still incredibly accurate. Quantum accuracy mesoscale simulations are now possible. orbitalmaterials.com/post/orb-v3-at…

#Perovskite_Power_Hour🤝 Join us tomorrow for 6⃣th edition of #ViPerCon2025💡Dive into the #past,#present & #future of perovskite with Laboratory of Photonics and Interfaces Maria Antonietta Loi Vaynzof lab Steve Albrecht's Lab⚛️ 📅15 Apr🕤09:50 Last chance to register👇 salibalabs.com/vipercon2025/ Michael Saliba Uni Stuttgart Forschungszentrum Jülich | @[email protected]

𝘈𝘵 𝘵𝘪𝘮𝘦𝘴, 𝘪𝘵’𝘴 𝘵𝘩𝘦 𝘴𝘭𝘰𝘸 𝘫𝘰𝘶𝘳𝘯𝘦𝘺𝘴 𝘵𝘩𝘢𝘵 𝘭𝘦𝘢𝘥 𝘶𝘴 𝘵𝘰 𝘵𝘩𝘦 𝘮𝘰𝘴𝘵 𝘱𝘳𝘰𝘧𝘰𝘶𝘯𝘥 𝘢𝘯𝘥 𝘣𝘦𝘢𝘶𝘵𝘪𝘧𝘶𝘭 𝘷𝘪𝘦𝘸𝘴. New in 𝐒𝐦𝐚𝐥𝐥 Wiley in research Wiley Chemistry : onlinelibrary.wiley.com/doi/10.1002/sm…

Now online: News & Views on the Article by Meihui Song, Bo Zhao, Yunlong Li, Feng Yang & co-workers nature.com/articles/s4416… ($) Encapsulated halide perovskites by Jeremy Sloan nature.com/articles/s4416… ($)

We have released an eSEN model that is the current SOTA on Matbench-Discovery. Code/checkpoints are available for both non-commercial and commercial use: code: github.com/facebookresear… checkpoint: huggingface.co/facebook/OMAT24 paper (updated): arxiv.org/abs/2502.12147

🚨 Thrilled to share Efficient Geometric Interatomic Potential (EGIP) - an efficient and accurate ML model for materials! Even though non-conservative MLIPs are not physically grounded, they offer significant efficiency. This work demonstrates that they are quite accurate on

On K_SRME metric, we find that graph construction, precision, and displacements play a key role in accurate predictions. If we just increase the displacements to 0.05 A , we can use much faster models.

PSA for machine learning force field authors who already have or are planning to submit models to Matbench Discovery. There's a new dynamic scatter plot on the landing page (…tbench-discovery.materialsproject.org) that allows comparing all 30 existing models against each other across 2 dozen

NEP89 is a foundation model based on the neuroevolution potential (NEP) architecture, delivering empirical-potential-like speed and near-DFT accuracy for large-scale atomistic simulations of inorganic and organic systems across 89 chemical elements. arxiv.org/abs/2504.21286 🚀

One issue with academic career is that it focuses on results rather than on understanding, with the assumption that getting results entails understanding, which is not always the case. So we may end up examining CVs of people who got many results but who understand almost nothing

We will soon have an opening in our group as part of a joint PhD program between NTU and Sorbonne University (36 months in SG, 12 months in Paris). Start date can be either in Jan or Aug 2026. Reach out if you're interested or know prospective candidates!

here's what the symmetry mismatch Sankey diagrams look like: …tbench-discovery.materialsproject.org/tasks/geo-opt#… another thing i've thought about plotting with Sankey diagrams is DFT vs MLIP stability mismatch, taking into account both thermodynamic and kinetic stability which gives 4 options: 1. on convex

Excited to share our latest releases to the FAIR Chemistry’s family of open datasets and models: OMol25 and UMA! AI at Meta FAIR Chemistry OMol25: huggingface.co/facebook/OMol25 UMA: huggingface.co/facebook/UMA Blog: ai.meta.com/blog/meta-fair… Demo: huggingface.co/spaces/faceboo…

✨💥 Our new art for a #Nature_paper that takes a major step toward real-world quantum technologies! The research, led by an incredible team including Melike Biliroglu and Mustafa Türe, and senior author Kenan Gundogdu, uncovers how #superfluorescence (an extremely rare and

Very excited and proud of this incredible team effort #compchem

Matbench Discovery is out in Nature Machine Intelligence @ Paper: rdcu.be/esRFB Leaderboard: …tbench-discovery.materialsproject.org No better time to thank all my co-authors Rhys Goodall, Philipp Benner, Yuan Chiang Yuan (Cyrus) Chiang, Bowen Deng, Mark Asta, Gerbrand Ceder Ceder Group,

Happy to introduce AlphaGenome, Google DeepMind's new AI model for genomics. AlphaGenome offers a comprehensive view of the human non-coding genome by predicting the impact of DNA variations. It will deepen our understanding of disease biology and open new avenues of research.

Last call. We will close the application to the following 2 jobs on 2025/11/03. Senior Researcher in Deep Learning: jobs.careers.microsoft.com/global/en/job/… Senior Research Engineer on Data: jobs.careers.microsoft.com/global/en/job/…