According to Odaily, OpenAI has launched a new feature allowing enterprise customers to customize its most powerful model, GPT-4, using their own company data. This move comes as startups face increasing competition in AI products, and enterprises are under growing pressure to demonstrate returns on AI investments. OpenAI plans to introduce this customization feature, commonly referred to as fine-tuning in the AI industry, on Tuesday. Fine-tuning enables existing AI models to be trained with additional information on specific tasks or subject areas.
This feature is a new addition to OpenAI's flagship model. Olivier Godement, OpenAI's API Product Lead, stated that the company aims to make it easier for customers to adjust its most powerful model by working directly with enterprises, rather than relying on external services or less capable products. To fine-tune the model, customers must upload their data to OpenAI's servers. John Allard, a software engineer at OpenAI involved in the customization work, mentioned that training typically takes one to two hours. Initially, users can only fine-tune the model using text-based data, with image or other content types not yet supported.
