You can now fine-tune your enterprise’s own version of OpenAI’s o4-mini reasoning model with reinforcement learning
1 min read
Summary
OpenAI’s new Reinforcement Fine-Tuning (RFT) tool enables third-party developers to customize and adapt its o4-mini AI language model for enterprise use.
It allows developers to create a private, enterprise-specific version of the model that can be tuned to more accurately mirror an organization’s internal terminology, objectives, and unique product language.
OpenAI has provided examples of early customer use cases, including Accordance AI, which used RFT to improve tax analysis tasks by 39%, and SafetyKit, which used the tool to increase its content moderation policy enforcement to 90%.
RFT is available to verified organizations through OpenAI’s online developer platform, with a 50% discount offered to teams that share their training data.
Pricing is based on the time spent on active training, with an hourly rate of $100, prorated to the second.