Google DeepMind has released AlphaGenome 2, a gene prediction model that analyzes DNA sequences of approximately 1 million bases to predict genomic function across specified tissues, cell types, or states. The model generates readouts covering gene expression, histone modification, chromatin accessibility, and newly added DNA methylation. Unlike its predecessor, which predicted multiple tissues in parallel, AlphaGenome 2 processes a single condition at a time, enabling it to distinguish between similar cell types and generalize to new single-cell data without retraining. DeepMind reports the model identified nearly twice as many variants affecting cellular gene expression compared to the previous state-of-the-art system. AlphaGenome 2 is currently available for non-commercial research via the AlphaGenome API, with commercial access in private beta. Related Skills are slated to launch on Google's agent development platform Antigravity in the near future.