Boltz, Excelra Team Up on AI Drug Discovery
The partnership pairs Boltz's AI models with Excelra's curated GOSTAR chemistry database to support early drug candidate screening.
Updated on Nov. 10, 2026 in Biotech

Boltz PBC announced a partnership with Excelra to train its BoltzMol-1 AI drug discovery models on curated ADME measurements from Excelra's GOSTAR database. The announcement was made on November 10, 2026.
Why it matters
Early-stage drug discovery depends on filtering out compounds with poor solubility, permeability or other unfavorable properties before costly synthesis. Boltz reported that its models identified binders for six of ten challenging targets in prospective testing, a result it presents as evidence the approach can work beyond its training data.
Boltz reports confirmed binders or active compounds for six of ten targets, using 28-96 compounds per target, versus conventional high-throughput screen hit rates often below 0.15%. These figures come from the company's own technical report, and independent validation has not been reported.
The players
Boltz PBC
Boltz PBC is an applied AI research lab and public benefit corporation that develops the BoltzMol-1 drug discovery models.
Excelra
Excelra is a company that operates the GOSTAR database, a curated chemical and pharmacological data resource, and is based in Hyderabad, India.
The details
Excelra's GOSTAR database, which offers a curated, ISO 9001-certified view of over 11 million small molecules, supplies measurements of kinetic solubility, lipophilicity and Caco-2 permeability used to train BoltzMol-1's ADMET models. Boltz says these models act as an early triage layer, and that its GOSTAR-trained ADMET models matched or exceeded a leading open-source package on out-of-distribution chemistry, with the largest gains in solubility and permeability.
Timeline
• November 10, 2026: Boltz and Excelra partnership announced via PRNewswire
The Tech Race
Conventional high-throughput screening has long been the standard first step in finding drug candidates, with hit rates often below 0.15%. Boltz's claim of active compounds for six of ten targets positions AI-driven triage as a competing method in the broader race to shorten early-stage drug discovery.
For patients, AI-driven candidate screening could eventually shorten the time before promising medicines reach clinical testing, though no timeline is given in the announcement. The results so far are company-reported and do not yet change any treatment or product available to consumers.
The takeaway
AI triage of compounds before physical screening may become a practical way to cut the cost and time of early drug discovery. Readers should treat the reported hit rates as preliminary until independent groups reproduce them.
Further reading
Explore more Biotech coverage for context on how AI is reshaping early drug discovery.
More information
Learn more on the Excelra GOSTAR database information page.



