By Milcah Tanimu
OpenAI has launched a new feature enabling businesses and developers to fine-tune its advanced GPT-4o model using their own data for $25, starting September 24, 2024. This enhancement allows for tailored adjustments in the model’s responses, accommodating specific needs such as tone, structure, and complex domain instructions.
In a recent blog post, OpenAI revealed that the fine-tuning option is now fully operational. Olivier Godement, OpenAI’s API head of product, emphasized the company’s commitment to simplifying the fine-tuning process, stating, “We’ve been extremely focused on lowering the bar, the friction, the amount of work it takes to get started.”
**Pricing and Availability**
The fine-tuning feature is accessible to all paid plan users. To begin, users can access the fine-tuning dashboard, select the base model, and start customizing GPT-4o. Until September 23, 2024, OpenAI offers 1 million free training tokens daily per organization. Post this period, training will cost $25 per million tokens, with inference priced at $3.75 per million input tokens and $15 per million output tokens.
For a more affordable option, the GPT-4o mini fine-tuning includes 2 million free training tokens daily during the same timeframe.
Key Benefits of GPT-4o Fine-Tuning
– Enhances the model’s ability to manage structure, tone, and complex instructions for specialized applications.
– Customization typically requires only one to two hours, enabling quick deployment.
– Includes robust safety measures and automated evaluations to prevent misuse.
Implications
Businesses can now fine-tune GPT-4o to create specialized tools such as customer service chatbots capable of handling detailed inquiries with minimal training data. This new feature, highly anticipated by developers, represents a significant leap in AI customization.
Data Privacy and Safety
OpenAI ensures that fine-tuned models remain under the user’s control, with full ownership of business data. The company has incorporated multiple safety protocols, including ongoing automated evaluations, to ensure compliance with usage policies and prevent misuse.
