August 21st.OpenAI Introduced a new feature that allows enterprise customers to use their own data to customize their most powerful AI models GPT-4oThe move is designed to address the growing competition for AI enterprise applications and meet the increased demand for return on AI investment. The move is designed to address the growing competition for AI enterprise applications and to meet the higher demand for return on AI investment from organizations.
The so-called customization is known in the industry as "fine-tuning".Existing AI models can be optimized for specific tasks or domains through fine-tuning. For example, a skateboarding company could fine-tune an AI model to become a customer service chatbot capable of answering questions about wheel and skateboard care.
Previously, OpenAI only allowed users to fine-tune its smaller models, such as the GPT-4o mini. opening up the fine-tuning capabilities of the GPT-4o and GPT-4 means that companies can more directly optimize OpenAI's most powerful models without having to rely on an external service or use a lower-performing model, said Olivier Godement, Head of Product at OpenAI. Olivier Godement, Head of Product at OpenAI, said, "We want to make it easier for customers to tune their most powerful models and streamline their operations by working directly with companies.
It is understood that companies need to upload data to OpenAI servers for model fine-tuning, which takes one to two hours on average. Currently.Fine-tuning only supports text-based data, other types of data such as images are not supported at this time.