OpenAI Announces New Features for AI Models During DevDay Event

OpenAI revealed yesterday new tools for developers to build applications with its AI models during Tuesday’s DevDay event in San Francisco.

According to Inc., the startup shared four major API (application programming interface) updates: RealTime, Prompt Caching, Model Distillation, and Vision Fine-Tuning. The new features will allow developers to use OpenAI’s new technology and apply it to their own products.

The newest and most innovative API service is RealTime which will allow developers to take advantage of the recently launched voice feature for ChatGPT. The new feature will process audio immediately—it was possible before, but required transcriptions and connecting different applications—and will be faster and cheaper. RealTime is expected to also handle video in the future.

Prompt Caching allows developers to reuse common prompts to reduce costs and work more efficiently. “This can reduce latency by up to 80% and cost by 50% for long prompts,” explained OpenAI on its platform for developers.

With Model Distillation developers will be able to expand and improve the capabilities of smaller AI models like GPT-4o mini, and Vision Fine-Tuning will help applications that rely on images to improve object detection, and process visual content better.

OpenAI shared on its website and developers’ platform more details, examples—including videos and interviews with other companies using the features—, and pricing of the new features. For RealTime, for example, the rates will be around $0.06 per minute for audio input and $0.24 for audio output.

This new announcement comes just days after Mira Murati’s—OpenAI’s Chief Technology Officer—and other executives resigning.