OpenEvidence Launches Four Medical AI Models for Clinicians

Three models-Osler, Sackett and Snow are available free to verified US clinicians through the OpenEvidence website and its iOS and Android apps, while Darwin is being offered through a limited research preview.

OpenEvidence has launched four medical AI models designed to give clinicians different options for rapid clinical answers, deeper evidence reviews and advanced medical research. Three models-Osler, Sackett and Snow are available free to verified US clinicians through the OpenEvidence website and its iOS and Android apps, while Darwin is being offered through a limited research preview.

Osler is designed for questions that arise during patient encounters and takes about five seconds to generate a response. It replaces the model previously used to generate OpenEvidence answers and will serve as the platform’s default model.

Sackett takes roughly 30 seconds to respond and is intended for clinical questions requiring a more detailed assessment of available evidence. The model can request additional clinical context before evaluating relevant medical research.

Snow, the successor to OpenEvidence Deep Consult, can take around five minutes to produce a response. According to the company, it examines multiple lines of inquiry across medical literature before generating a detailed report.

OpenEvidence said the three production models follow the same clinical-accuracy standards, with differences primarily based on response time and the depth of literature searches.

Darwin is positioned as the company’s most advanced model and will initially be restricted to institutional partners, research collaborators and accredited academic AI researchers. OpenEvidence said Darwin achieved a 100% score on MedQA, a benchmark based on medical licensing examination questions. The company is also publishing the model’s responses and explanations for each answer.

The model scored 72.8% on MedXpertQA, 82.7% on HealthBench Professional and 87.2% on NOHARM, according to OpenEvidence. The company said Darwin outperformed competing models evaluated on the benchmarks.

OpenEvidence is limiting Darwin’s access because of potential dual-use risks linked to advanced medical and scientific reasoning, including possible misuse in virology, bioweapons-related research and human germline editing.

Initial Darwin users include the National Organization for Rare Disorders, which plans to use the model for complex cases. OpenEvidence said capabilities tested through the research preview could eventually be incorporated into Osler, Sackett and Snow as safeguards are validated.

The launch follows OpenEvidence’s $250 million funding round in January at a $12 billion valuation, bringing total funding to nearly $700 million. In July, the company partnered with Boston Children’s Hospital to integrate its AI into the EHR and expanded access across NewYork-Presbyterian, Columbia and Weill Cornell Medicine hospitals and care sites in New York City and Westchester.

Stay tuned for more such updates on Digital Health News