Google DeepMind Releases AlphaGenome 2 Genome Model
Understand the story
On November 11, 2026, Google DeepMind released AlphaGenome 2, its next-generation DNA sequence model, which can predict the effects of DNA sequences and their variants on gene regulation across hundreds of human cell types. The model adopts a conditional architecture, and its training data is based on over 107 million single-cell profiles covering 68 tissue types; compared with the earlier reported phrasing of "more than 100 million cell profiles," the official first-hand information gives the specific figure of 107 million. In evaluations on the TenK10K dataset, the number of causal variants identified by the model is nearly twice that of previous models, with a ratio of 30% versus 16%, and it can predict new cell types and biological states without retraining. The model is now open to non-commercial research via the AlphaGenome API, with limited private preview available for commercial research. The event is still at the model release stage, with DeepMind having published the architecture changes, training data, and evaluation results.
AI overview of reporting · Updated 2 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- Google DeepMindAlphaGenome 2: Deciphering how cells read DNA to better understand disease
Google DeepMind released AlphaGenome 2, a next-generation DNA sequence model that predicts how DNA sequences and their variants affect gene regulation across hundreds of human cell types. The model uses a conditional architecture and was trained on over 107 million single-cell profiles spanning 68 tissues; on the TenK10K dataset it identifies nearly twice as many causal variants as the previous model (30% vs 16%).
- The Verge · AIGoogle’s AlphaGenome 2 zooms in on the cellular roots of disease
Google DeepMind released the genomic prediction model AlphaGenome 2, which can predict how genetic variants affect individual cell types and states. The model was trained on more than 100 million cell profiles from 68 tissues and identifies about twice as many variants that alter gene expression as its predecessor, and can predict new cell types and biological states without retraining. It is available via the AlphaGenome API for non-commercial research, with limited private preview for commercial research.
Story heat trend
Not enough continuous observations to show a trend yet.