1. OpenAI is expanding its Custom Model program to help enterprise customers develop tailored generative AI models for specific use cases.
2. The program now includes assisted fine-tuning and custom-trained models to maximize performance and allow organizations to build personalized models.
3. The new features for developers working with GPT-3.5 include a dashboard for comparing model quality, support for integrations with third-party platforms, and enhancements to tooling, but details on fine-tuning for GPT-4 are not yet released.
OpenAI has announced the expansion of their Custom Model program, which helps enterprise customers develop tailored generative AI models for specific use cases, domains, and applications. The program was launched last year at OpenAI’s developer conference and has seen dozens of customers enroll since then. The expansion includes new features such as assisted fine-tuning and custom-trained models.
Assisted fine-tuning goes beyond traditional fine-tuning techniques to help organizations set up data training pipelines and evaluation systems to improve model performance on specific tasks. Custom-trained models are built with OpenAI’s base models and tools, such as GPT-4, for customers who need to deeply fine-tune their models or incorporate new, domain-specific knowledge.
Examples of successful custom models include SK Telecom using GPT-4 to improve telecom-related conversations and Harvey creating a custom model for case law. OpenAI believes that customized models personalized to specific industries or use cases will become more common in the future, allowing organizations of all sizes to realize more meaningful impacts from their AI implementations.
The expansion of the Custom Model program aligns with OpenAI’s goal of maintaining revenue growth as they reportedly approach $2 billion in annualized revenue. Custom models can also reduce strain on OpenAI’s model serving infrastructure by being more performant and smaller than general-purpose models. Additionally, OpenAI has unveiled new features for developers working with GPT-3.5, including a dashboard for comparing model quality, support for integrations with third-party platforms, and enhancements to tooling.