Six Specificity Foundation Models (SFMs) across six molecular-recognition domains — TF–DNA, enzyme–substrate, peptide–MHC, CRISPR off-target, miRNA–mRNA, and drug–target — built by vibe coding from one physics-derived, sequence-native architecture (Reddy 2026).

SupraBench: the first benchmark for evaluating LLMs on supramolecular host-guest chemistry reasoning (4 scored tasks + a vision task + a domain corpus).

1st-place solution toolkit (1/45) for Bohrium Scientific Data Annotation Competition Track 1: molecular images to Extended-SMILES.