Iambic has launched Enchant v3, the latest generation of its multimodal transformer model for drug discovery and development. The company introduced Enchant v3 as part of its molecular superintelligence platform. The model supports decisions across the end-to-end drug discovery and development process.
Iambic is a clinical-stage life science and technology company developing novel medicines through its AI-driven platform. Enchant serves as the ‘brain’ behind the company’s molecular superintelligence platform. The platform works with Iambic’s in-house high-throughput experimental chemistry and biology capabilities. Together, these technologies can learn and predict multiple properties for a target compound profile.
According to Iambic, Enchant v3 uses 41 billion parameters. The model also draws from more than 6,000 molecular properties across 16 biomedical data modalities. Furthermore, Enchant v3 can process diverse datasets from multiple sources. These include public and proprietary data across several scientific modalities. The model also uses a mixture-of-experts architecture. In addition, it evaluates multiple hypotheses in parallel. This approach helps optimize decisions throughout drug discovery and development.
“Enchant v3 is an architectural leap over Enchant v2, with new capabilities to be deployed for internal and partner drug discovery and development efforts,” said Fred Manby, PhD, Co-Founder and CTO, Iambic. “Earlier versions of Enchant have demonstrated utility across our pipeline, helping us predict critical preclinical and clinical endpoints. Enchant has shown robust scaling laws, with larger models exhibiting more powerful predictive capabilities. Enchant v3 continues this trend and constitutes our next step to make better technology for better medicines.”
Enchant v3 Expands Multimodal Drug Discovery
Enchant was developed to work with a broad range of biomedical information. The model can ingest assays, text, images, multi-omics, and biomolecular structures. It can also process biological sequences and clinical trial data. Moreover, it can combine public information with proprietary datasets. The platform uses model scale to improve prediction performance across related endpoints. It can also learn from data involving different molecules and related properties.
Additionally, Enchant aims to break down data barriers between preclinical research and clinical development. This approach can help connect information across different stages of the drug development process. The model also converts its outputs into probability-guided decisions. It uses uncertainty quantification to help determine which laboratory experiments can provide useful information. As a result, researchers can prioritize potential compounds using model-driven insights. This process supports multi-parameter optimization during drug discovery.
Enchant v3 also introduces several major technical upgrades. The model contains 41 billion parameters, supporting greater reasoning complexity. It was pretrained on 4.5 trillion tokens. In addition, it covers 6,000+ molecular properties for multi-parameter optimization. The model supports 16 biomedical modalities. These include text, assay data, molecule structure, protein structure, and 3D structure. It also covers images, multi-omics, biologics, clinical trial data, and pharmacokinetics. Therefore, Enchant v3 can examine chemistry, biology, and clinical tractability together.
Iambic Advances Molecular AI Capabilities
Enchant v3 also demonstrates predictable scaling laws, according to Iambic. The company said greater model scale correlates with stronger prediction accuracy. These improvements span multiple preclinical and clinical endpoints. The findings build on three generations of the Enchant model.
“When we launched Enchant in 2024, it was not obvious that scaling laws would hold in a domain like drug discovery,” said Matt Welborn, PhD, SVP of Machine Learning, Iambic. “Biomedical data is richer than text, more complicated to work with, but allows us to build a unique capability. Across three generations of the model, the scaling has held – whenever we have added parameters, data, and modalities, the predictions improved, including for endpoints where data is sparse.”
With Enchant v3, Iambic aims to support two key objectives. First, the company wants to increase the probability of success for drug candidates. Second, Iambic plans to expand into new therapeutic areas and modalities. This expansion is intended to advance additional internal and partner programs.
The company also highlighted the role of Enchant in its broader drug discovery pipeline. For greater insight, including specific examples of how Enchant has helped advance the Iambic pipeline
Explore Health Tech Insiders for the latest medical innovations and reliable strategic insights driving the future of technology-driven healthcare transformation.
News Source: Businesswire.com
