AI models aimed at life sciences

Drug discovery starts with guesses about biology, then years of lab work to test them. OpenAI and Amazon Web Services both launched new AI products this week to help generate those guesses earlier in the process. (openai.com; aboutamazon.com) OpenAI said on April 16 it introduced GPT‑Rosalind, a life-sciences model built for biology, drug discovery and translational medicine, with access through ChatGPT, Codex and the application programming interface for qualified customers. Reuters reported the model is named after Rosalind Franklin and is being offered as a research preview under OpenAI’s trusted-access structure. (openai.com; indianexpress.com) Amazon Web Services said on April 14 that Amazon Bio Discovery is now generally available as a single application that combines biological foundation models, agentic assistants and contract research organization lab workflows. Amazon said the product gives scientists access to more than 40 biology models and lets wet-lab results flow back into the next design cycle. (aws.amazon.com; aboutamazon.com) A biological foundation model is software trained on large stores of sequence and lab data, the way a language model is trained on text. Companies use those models to predict which antibodies, proteins or molecules are worth testing before they spend money making them in a lab. (aboutamazon.com; aws.amazon.com) OpenAI framed GPT‑Rosalind as a reasoning model for scientific workflows, saying it is optimized for chemistry, protein engineering and genomics. The company said a new drug in the United States typically takes about 10 to 15 years from target discovery to regulatory approval, so earlier decisions carry through the rest of development. (openai.com) Amazon Web Services is pitching a slightly different layer of the stack: not just model output, but the handoff from software to physical testing. Its product page says researchers can generate candidates, send them to lab partners for synthesis and testing, and use the returned data to refine the next round. (aws.amazon.com; biodiscovery.aws.com) That puts both launches upstream of the chatbot use cases most people know. Instead of drafting emails or code, these tools are being sold to help decide which experiments to run, which molecules to prioritize and which hypotheses deserve scarce lab time. (openai.com; aboutamazon.com) Drugmakers and biotech companies have been buying into that idea for several years, but many projects still depend on fragmented software, manual handoffs and separate vendors for computation and wet-lab validation. Amazon Web Services said those handoffs slow the feedback loop and make experiments harder to reproduce across teams. (aws.amazon.com) OpenAI said GPT‑Rosalind includes stronger tool use and database use for multi-step scientific work, but it is limiting access to qualified users rather than opening the model broadly. Amazon Web Services, by contrast, is selling a managed application with built-in model selection and outside lab partners, which makes it closer to research infrastructure than a standalone model. (help.openai.com; aws.amazon.com) The near-term test is not whether these systems can write plausible biology. It is whether they help researchers pick better experiments, get cleaner validation data and cut down the number of dead ends before a program reaches the lab bench. (openai.com; aboutamazon.com)