OpenAI has introduced Reinforcement Fine-Tuning (RFT), a new model customization technique for its o1 and o1-mini models. This method allows developers to tailor AI models for domain-specific tasks by training them on datasets ranging from dozens to thousands of high-quality tasks and evaluating responses against reference answers. RFT enables the creation of expert models in fields like coding, scientific research, and finance, with the capability to outperform traditional fine-tuning methods by requiring only a handful of examples for efficient adaptation. A demonstration with Berkeley Lab showcased RFT's capabilities in rare disease gene prediction, where the fine-tuned o1-mini model achieved 31% accuracy, outperforming both the base o1-mini and standard o1. The preview of RFT is set to launch in early 2025, with applications now open for early access.