Flow-of-Options (FoO) empowers large language models to overcome inherent biases by systematically exploring diverse reasoning possibilities before arriving at solutions.

• Achieved 38.2% - 69.2% improvement on standard data science tasks
• Demonstrated 37.4% - 47.9% gains on therapeutic chemistry applications
• Enables autonomous problem-solving through deliberate consideration of multiple options

For medical applications, FoO's improved reasoning could enhance drug discovery processes, treatment planning systems, and diagnostic support tools—where considering diverse options is crucial for patient safety and effectiveness.

Flow-of-Options: Diversified and Improved LLM Reasoning by Thinking Through Options