Enki 4 re-architects the platform to deliver faster performance and broaden target and modality coverage, including degraders, PROTACs, glues, and novel payload designs for antibody drug conjugates and related modalities.

For more information and the full list of targets, readers can visit variational.ai/enki or contact [email protected].

The full list includes 760 targets, with readers invited to visit variational.ai/enki or reach out at the provided contact.

Variational AI released Enki 4, a major update to its generative AI platform for small-molecule drug discovery, expanding performance, target coverage, and modality support.

Enki 4 re-architects the platform and improves underlying algorithms to deliver faster performance and broader applicability, including degraders, PROTACs, glues, and antibody-drug conjugate payload design.

Company leadership describes Enki 4 as enabling biotech partners to begin from novel leads not readily discovered with traditional methods, bypassing hit identification and hit-to-lead steps for direct progression to lead optimization.

The company emphasizes ongoing model innovation and faster, more capable generation to support its partners in drug discovery.

Enki 4 helps biotech partners start with novel leads not readily found via traditional methods, aiming to bypass hit identification and hit-to-lead bottlenecks and produce molecules ready for lead optimization.

The update enables rapid generation of novel, potent, selective, and synthesizable lead-like structures pre-trained on 760 drug targets, up from 592 (a 28% increase), with more targets to be released soon.

The platform now rapidly generates novel lead-like structures that are potent, selective, and synthesizable, pre-trained on 760 targets, with ongoing target additions.

Enki is developed by Variational AI, a Vancouver-based company founded by researchers from MIT, Caltech, Google Research, Microsoft Research, and D-Wave Quantum, using state-of-the-art generative AI to design optimized small molecules.

Enki 4 improves performance across more than a dozen target classes, including GPCRs, kinases, ion channels, proteases, oxidoreductases, and hydrolases.