Insilico Medicine Launches Industry's First Drug Discovery and Development Benchmark-as-a-Service to Validate AI Models Against Real-World Pharmaceutical Data
Key Insights
Insilico Medicine (search) announced the Drug Discovery and Development (DDD) Benchmarks as a Service, the first standardized evaluation framework for measuring how frontier AI models perform on real-world drug discovery tasks.
The DDD Benchmark comprises two suites: Drug Discovery Foundations with over 300 evaluations, and Drug Candidate Essentials testing end-to-end drug discovery program navigation from hit identification to preclinical candidate nomination.
The framework is anchored in Insilico's proprietary validated programs, including 31 preclinical candidates nominated in six years and a Phase III AI-discovered TNIK (search) inhibitor, Rentosertib (search), for IPF.
As general-purpose and domain-specific AI models rapidly infiltrate drug discovery, one critical question looms over the pharmaceutical sector: Can these models actually discover drugs, or are they just good at taking tests? Today, most publicly available AI benchmarks suffer from severe data contamination, where models achieve artificially inflated scores simply by memorizing test questions present in their training data, only to fail when faced with real-world drug candidate decisions.
To close that gap, Insilico Medicine (search), a clinical-stage generative artificial intelligence-driven drug discovery company listed on the Hong Kong Stock Exchange (HKEX:3696), today announced the launch of the Drug Discovery and Development (DDD) Benchmarks as a Service (BaaS). This pioneering standardized evaluation framework is anchored in carefully decontaminated real-world data and proprietary validated programs, designed to measure how frontier AI and foundation models built on other architectures perform on real-world tasks across medicinal chemistry, chemical synthesis, disease biology, clinical development, and longevity research.
"The rapid progress of AI has made one question more urgent than ever: can these models actually discover drugs?" said Alex Zhavoronkov, Ph.D., founder and CEO of Insilico Medicine (search). "For more than a decade, we have built and validated generative AI across the entire drug discovery value chain, nominating 31 preclinical candidates in six years and advancing an AI-discovered and AI-designed medicine into Phase III. The DDD Benchmark converts that real-world experience into a rigorous, standardized yardstick that the whole field can use."
A Two-Suite Evaluation Architecture
The DDD Benchmark comprises two complementary evaluation suites, each targeting distinct aspects of AI-driven pharmaceutical research.
The first, Drug Discovery Foundations, includes more than 300 evaluations built from proprietary out-of-distribution test sets and rigorously decontaminated public data. It measures a model's core competencies across the discipline, spanning disease biology, molecular property prediction and optimization, retrosynthesis, structure-based design, and clinical development.
The second suite, Drug Candidate Essentials, evaluates a model's ability to navigate an end-to-end drug discovery program, from hit identification through preclinical candidate (PCC) nomination. Its reference baselines are anchored in Insilico's own validated programs, testing whether a model can make the sequential, high-stakes decisions that real drug discovery demands.
Accessible and Transparent Benchmarking
Benchmarking is designed for accessibility: any model served through a standard chat-completions API can be evaluated by contacting Insilico Medicine (search). The company scores the resulting outputs against expert reference baselines and provides a standardized scorecard comparing performance with leading models. Results are delivered with a verified score report that lets organizations certify performance for internal and partner audiences. Organizations seeking public recognition can also publish their results on the DDD Benchmark public leaderboard, giving the field a transparent, like-for-like view of model capability.
As AI systems increasingly act as agents — planning experiments, reasoning over experimental data, and calling MCP-compatible tools — the DDD Benchmark is designed to measure whether those capabilities translate into sound, real-world drug discovery decisions rather than strong performance on abstract tasks.
Built on a Decade of Validated Drug Discovery
The DDD Benchmark builds on Insilico's Pharma.AI platform and the MMAI Gym, the company's post-training environment for scientific AI. Since its founding, Insilico has nominated 31 preclinical candidates, received over 10 investigational new drug (IND) clearances, and compressed the timeline to PCC nomination to roughly 12 to 18 months, compared with the 2.5 to 4-plus years typical of traditional drug discovery.
Its lead program, Rentosertib (search) (ISM001-055), is a first-in-class, AI-discovered and AI-designed TNIK (search) inhibitor now in Phase III development for idiopathic pulmonary fibrosis (search) (IPF). This program serves as a tangible validation of the AI-driven discovery approach that underpins the new benchmarking framework.
"As we advance toward pharmaceutical superintelligence, measuring genuine capability, not memorized answers, is how we ensure AI delivers the highest-quality, differentiated medicines and helps extend healthy, productive longevity for people everywhere," Zhavoronkov added.
The DDD Benchmark as a Service is available now. Organizations interested in evaluating a model, requesting a private assessment, or placing a model on the public leaderboard can learn more at dddbench.insilico.com.