Logos: An evolvable reasoning engine for rational molecular design

Researchers introduce Logos, a compact AI model that combines multi-step logical reasoning with chemical consistency for molecular design. The model achieves strong performance in structural accuracy and chemical validity while using fewer parameters than larger language models, and provides transparent reasoning that can be inspected by humans.

• →Logos integrates multi-step logical reasoning with strict chemical consistency for molecular design tasks.
• →The model matches or surpasses larger general-purpose language models while using a fraction of their parameters.
• →Training involves staged exposure to reasoning examples, followed by alignment with molecular representations and chemical rule incorporation.
• →The system provides transparent, human-inspectable reasoning steps for each generated molecular structure.
• →Results demonstrate stable performance in molecular optimization tasks with multiple conflicting constraints.