This research demonstrates how optimized context retrieval can enhance open-source LLMs to match or exceed proprietary models for healthcare applications while dramatically reducing costs.

• Achieves state-of-the-art accuracy on medical question answering benchmarks
• Creates a significantly improved cost-accuracy Pareto frontier on MedQA
• Introduces OpenMedQA, a new benchmark for open-ended medical question answering
• Develops affordable and reliable LLM solutions specifically for healthcare contexts

This matters because healthcare organizations can now implement high-quality AI solutions without the prohibitive costs of proprietary models, democratizing access to advanced medical AI capabilities.

Pareto-Optimized Open-Source LLMs for Healthcare via Context Retrieval