Darwin's Our Take: OpenAI unveils GPT-Rosalind to support life sciences research

OpenAI unveils GPT-Rosalind to support life sciences research

ChatGPT creator OpenAI introduced a new artificial intelligence model called GPT-Rosalind it developed specifically to “help researchers accelerate the early stages of [drug] discovery.”  

The model was developed to support research in the areas of biology, drug discovery, and translational medicine. It can help with literature reviews, data analysis, hypothesis generation, the planning of experiments, and other tasks within complex research workflows.

In addition to facilitating workflows, GPT-Rosalind can help scientists explore more possibilities, see connections that might otherwise go unnoticed, and form “better hypotheses sooner,” OpenAI stated in the announcement.  

Noting that it takes about 10 to 15 years, on average, for a new drug to go from target discovery to regulatory approval in the U.S., OpenAI said “gains made at the earliest stages of discovery compound downstream” and pay off in multiple ways.

“Over time, [advanced AI] systems could help life sciences organizations discover breakthroughs that wouldn’t otherwise be possible, with a much higher rate of success.”

Biopharma and research organizations such as Amgen, Moderna, Thermo Fisher Scientific, and the Allen Institute are collaborating with OpenAI to apply the new AI model across their workflows.

Sean Bruich, senior vice president of AI and data at Amgen, said: “The life sciences field demands precision at every step. The questions are highly complex, the data are highly unique, and the stakes are incredibly high. Our unique collaboration with OpenAI enables us to apply their most advanced capabilities and tools in new and innovative ways with the potential to accelerate how we deliver medicines to patients.

OpenAI said GPT-Rosalind is available to qualified enterprise customers as a research preview through ChatGPT, Codex, and the API.

The model is named after Rosalind Franklin, a British chemist, “whose rigorous research helped reveal the structure of DNA and laid foundations for modern molecular biology,” OpenAI noted.

OpenAI also recently introduced ChatGPT for Clinicians, a version of ChatGPT developed to free up clinicians’ time by supporting tasks such as documentation and medical research.  

The company said it is making ChatGPT for Clinicians free “for any verified physician, NP, PA, or pharmacist, starting in the U.S.”

OUR TAKE: Novo Nordisk announced a partnership with OpenAI earlier this month to integrate AI capabilities into processes ranging from drug discovery to commercial operations.

Although the press release did not specifically mention GPT-Rosalind, Novo Nordisk said the partnership would apply advanced AI capabilities to analyze complete datasets, identify promising drug candidates, and shorten the time it takes “to move from research to patient.”

Mike Doustdar, Novo Nordisk’s CEO, said, “Integrating AI in our everyday work gives us the ability to analyze datasets at a scale that was previously impossible, identify patterns we could not see, and test hypotheses faster than ever. This means discovering new therapies and bringing them to market faster than ever before.”

The company also said it would apply OpenAI’s capabilities to improve efficiency in areas such as manufacturing, supply chain, and distribution.

According to Novo Nordisk, the partnership with OpenAI “has been structured with strict data protection, governance, and human oversight to ensure ethical and compliant use.”

Merck, meanwhile, has chosen to partner with Google Cloud and will invest up to $1 billion in the partnership to deploy an agentic platform across the company’s R&D, manufacturing, commercial, and corporate functions, according to the press release.

The deployment will include Google Cloud’s “most sophisticated AI, including Gemini Enterprise,” Merck noted.

“AI agents and generative tools will help our teams around the world reimagine processes at scale and bring scientific breakthroughs to patients faster,” said Dave Williams, Merck’s chief information and digital officer.

While the use of these advanced AI models holds vast potential for accelerating life sciences research, implementing them also carries enormous risk for unintended harm, as this news article by Stanford University’s Human-Centered Artificial Intelligence (HAI) institute points out.

“Our intent isn’t to be overly sensationalized or use scare tactics about worst-case scenarios, but there are very real issues,” said Artem Trotsyuk, a fellow with the Stanford Center for Biomedical Ethics.

Among the issues raised in the article are the possibility that AI-powered biomedical research could be used in ways that violate personal privacy and security, intensify racial bias, and even result in new types of toxic agents or bioweapons.  

A team of scientists, ethicists, and researchers at Stanford, in collaboration with peers at Harvard Medical School, Cleveland Clinic, and other organizations, created an ethical framework to address these risks — which Trotsyuk hoped would be a starting point for researchers and policymakers.

The group has been creating a series of papers pertaining to risk mitigation in this area, one of which was published by Nature Machine Intelligence. The abstract is available here.

What else you need to know

Revolution Medicine’s daraxonrasib continued to generate considerable buzz at the American Association for Cancer Annual Meeting in Chicago. The investigational therapy is being evaluated in patients with pancreatic cancer as both a first- and second-line treatment, and as a monotherapy or in combination with chemotherapy. The once-daily oral drug is a multi-selective inhibitor of Ras proteins.

Earlier this month, Redwood City, Calif.-based RevMed released top-line results from a Phase III trial of daraxonrasib in patients with metastatic pancreatic ductal adenocarcinoma (PDAC) who had already received other treatment. As compared with standard-of-care IV chemotherapy, daraxonrasib demonstrated a significantly better and “unprecedented” rate of median overall survival (OS; 13.2 months vs. 6.7 months). Daraxonrasib met all of the study’s other primary and secondary endpoints, including progression-free survival (PFS).

Then last week, RevMed presented data from a pair of open-label Phase I/II trials of daraxonrasib in patients with previously untreated metastatic PDAC. In one trial, patients with Ras-mutant gastrointestinal tumors received daraxonrasib plus gemcitabine and nab-paclitaxel. In the other trial, patients with Ras-mutant solid tumors received daraxonrasib monotherapy.

Although median PFS and OS were not mature at the data cutoff in either trial, the available data showed an OS of 90% and a PSF rate of 84% at six months in the combination therapy trial, and an OS rate of 83% and a PSF rate of 71% at six months in the monotherapy trial.

Pancreatic cancer is the most Ras-addicted of all major cancers, RevMed noted, adding that more than 90% of patients have tumors driven by mutations in Ras proteins.    

Amazon One Medical launched a GLP-1 management program designed to integrate primary care, pharmacy services, and virtual care for those who use these drugs for weight loss. The announcement said the program would benefit patients by offering “a fully integrated care journey” in which clinicians can monitor patients’ progress, adjust treatments, and address potential comorbidities such as cardiovascular disease and diabetes.

At the same time, Amazon expanded access to GLP-1 drugs through Amazon Pharmacy, providing upfront pricing on oral and injectable versions. Those with insurance may pay as little as $25 per month, while costs for those paying cash could be as little as $149 per month for oral GLP-1s or $299 per month for injectable GLP-1s. Currently, same-day delivery is available in almost 3,000 cities, and the total is expected to reach nearly 4,500 by year-end.

Meanwhile, CMS is delaying part of the five-year pilot program called BALANCE, which was designed to expand access to GLP-1s for those enrolled in Medicare and Medicaid. Aetna and other insurers expressed doubts about participating in the program, prompting CMS to extend the Medicare “bridge” program that is slated to start in July for an extra year, through the end of 2027. Under that program, enrollees will pay a $50 copay and CMS will cover the balance of the negotiated prices. The agency plans to proceed with the Medicaid portion of the pilot program, which could begin in May.  

ChristianaCare and Cardiovascular Physicians of Delaware will open a new ambulatory surgery center in Newark, Del., through their recently formed joint venture with US Health Partners. The facility of nearly 9,000 square feet will be located on the second floor of the HealthCare Center at Wilmington, Del.-based ChristianaCare and will provide outpatient heart and vascular procedures such as diagnostic heart catheterizations, coronary and peripheral vascular interventions, ablations, and cardiac device implantations. Atlas Healthcare Partners will manage and operate the ASC, which is expected to open in late 2027, according to the announcement.  

Oscar Health has unveiled an AI-powered platform called Lucie Health

Marketplace, where consumers can shop for individual health plans among “all major medical carriers,” as well as ancillary and supplemental products such as dental and vision plans. Employers can also use the new platform to offer individual coverage health reimbursement arrangements (ICHRAs) to members. Oscar noted that Lucie includes individual networks in almost all ZIP codes. “Lucie instantly quotes, enrolls, and renews products across carriers,” Oscar said in the announcement, adding that brokers can sign clients up in minutes. Consumers can use a broker if they want assistance or enroll on their own.

Eli Lilly agreed to acquire Kelonia Therapeutics, a privately health biotech based in Boston, in a deal potentially worth $7 billion. Kelonia has developed a proprietary in vivo gene replacement system that uses specially engineered lentiviral-based particles to selectively enter and bind to T-cells within the body, where they deliver a gene that turns the T-cells into cancer-fighting CAR-T cells. Kelonia’s lead candidate is an investigational gene therapy that targets the BCMA protein expressed on the surface of multiple myeloma cells. Under the definitive agreement the companies signed, Lilly will pay $3.25 billion up front, with up to another $3.75 billion to be paid if certain clinical, regulatory, and commercial milestones are achieved. If customary closing conditions are met, the transaction is expected to close in the second half of this year.  

What we’re reading

Oncologists vs. the FDA and RFK Jr. WSJ, 4.21.26 (subscription or registration required)

Hospital at Home and Transforming US Health Care Delivery. JAMA, 4.23.26 (subscription or registration required)

New drugs are primary care game changers, but pricey. Medscape, 4.21.26 (registration required)

‍