Insilico Medicine launched what it calls the drug discovery benchmark-as-a-service for frontier AI models, targeting weaknesses in how AI systems are evaluated. The initiative is designed to test whether models can make meaningful scientific decisions instead of merely recalling patterns from training data. The benchmark is positioned as a “Drug Discovery and Development Benchmark as a Service,” intended to provide structured measurement for performance in drug discovery workflows—an area where many existing benchmarks have been criticized for limited real-world decision relevance. For the AI-in-biopharma market, the move signals a shift toward externalized validation frameworks as companies compete to deploy increasingly capable foundation models across target identification, lead optimization, and development planning. The launch also reflects investor and partner pressure for defensible AI claims, particularly when systems are marketed as accelerating preclinical discovery decisions rather than assisting only with text or knowledge retrieval.