Microsoft Research released Skala-1.1, a deep-learning DFT functional trained on 2.5× more data than the first public version, achieving a weighted average error of 2.8 kcal/mol on GMTKN55.
MindTopo is a new benchmark for testing topological reasoning in AI, evaluating whether multimodal models can understand concepts such as connectivity, enclosure, order, separation, and knots.
CARE-X is a research model from Microsoft Research that combines generative and discriminative capabilities for chest X-ray interpretation, using reinforcement learning (DAPO) and tool-augmented…
Microsoft Research released Orchard, an open-source framework for scalable agentic AI, including three domain-specific training recipes (Orchard-SWE, Orchard-GUI, Orchard-Claw) and associated…
Microsoft Research introduces Echoverse, a set of twelve synthetic training worlds for computer-use agents, designed with high fidelity and depth.
Microsoft Research introduces EvoLib, a framework that enables large language models to learn from their own experience during inference by transforming raw experiences into an evolving library of…
Microsoft Research announces the release of formally verified cryptographic code in SymCrypt, using Rust, Lean, Aeneas, and AI agents for proof automation, initially for SHA-3 and ML-KEM.
Microsoft Research announced Aurora 1.5, a major update to the open Aurora Earth-system foundation model, adding 22 new weather variables, hourly temporal resolution, and probabilistic ensemble…
Microsoft Research introduces Flint, a visualization intermediate language for AI-driven chart creation that compiles simple, human-editable specifications into polished charts across multiple…
SkillOpt is a method that treats agent skill files as trainable parameters outside a frozen model, using a training-style optimization loop to improve agent performance without changing model…
Memora is a scalable memory system for long-horizon AI agents that decouples rich memory content from lightweight retrieval, achieving state-of-the-art results on LoCoMo and LongMemEval benchmarks…
Microsoft Research, in collaboration with UC Berkeley, UCSF, and Columbia University, published a paper in Nature Neuroscience introducing generative causal testing (GCT), a framework that distills…
Talos is an open-source tool for automated, iterative reanalysis of genomic data to diagnose rare diseases, designed to be conservative with low false-positive rates, recovering 90% of diagnoses in…