Google's new AI model helps discover potential cancer-fighting drugs
What's the story
Google has announced a significant milestone in cancer therapy research, thanks to its advanced artificial intelligence (AI) model, C2S-Scale 27B. The foundation model was developed in collaboration with Yale University and works by understanding the "language" of individual cells. It has already predicted new drugs that could potentially fight tumors, based on its observations of cancer at a microscopic level.
Model features
C2S-Scale 27B now available for other researchers
The C2S-Scale 27B model, which is based on the Gemma family of open models, has been built with a whopping 27 billion parameters. Google has made this powerful tool available for other researchers on platforms like GitHub and Hugging Face. This move could potentially accelerate the search for effective cancer treatments using AI technology.
Drug discovery
Model tasked with identifying drug that amplifies immune signal
The C2S-Scale model was tasked with identifying a drug that could act as a conditional amplifier, boosting the body's immune signal only in a specific "immune-context-positive" environment. This required advanced reasoning capabilities, which were beyond those of smaller models. The researchers simulated over 4,000 drugs across different contexts and asked the model to predict which ones would only enhance antigen presentation in this positive immune context.
Twitter Post
Google CEO also made an announcement on X
An exciting milestone for AI in science: Our C2S-Scale 27B foundation model, built with @Yale and based on Gemma, generated a novel hypothesis about cancer cellular behavior, which scientists experimentally validated in living cells.
— Sundar Pichai (@sundarpichai) October 15, 2025
With more preclinical and clinical tests,…
Model predictions
Some predictions were already known, others were surprising discoveries
The model made several predictions, some of which were already known from previous studies, while others were surprising discoveries with no prior links. One such prediction was a significant increase in antigen presentation when the kinase CK2 inhibitor silmitasertib (CX-4945) was applied in the "immune-context-positive" setting. This finding highlights the potential of C2S-Scale 27B in identifying new drug candidates for cancer treatment.