these 4 OpenAI New AI Features would change the world

On a recent developer-facing event called Developer Day, OpenAI launched four new features powered by AI which are targeted to assist the needs of businesspersons and developers in creating AI products. These updates also provide more possibilities for customization, reduce costs and enhance the effectiveness of smaller models. So, let’s take a look at the features that are available in the OpenAI API at the moment:

1. Model Distillation

Model distillation was one of the new introductions from OpenAI where smaller models such as GPT 4o mini can internalize the learning obtained from bigger models. In previous times, this was a labor-intensive and delicate task where developers had to handle multiple tools at once. OpenAI’s solution makes this capability available in the API without requiring such integration. Thus, developers now can also create datasets with the help of advanced models, narrow down advanced models trained with that data, and domain-specific performance measurements are developed.

To help developers make this change, OpenAI is also offering two million free training tokens daily for GPT-4o mini and one million for GPT-4o till October 31. From this date onwards, standard pricing for the deployment of fine-tuning will be in effect.

2. Prompt Caching

In an effort to cut operational costs, OpenAI has introduced prompt caching. This feature allows developers to evade the need to pay the full amount required for every call by storing prompt prefixes for a maximum of one hour. OpenAI provides economies of scale as a new prompt that uses an already cached prefix attracts a 50% discount on the input cost. Thus it becomes very cheap for developers who are developing focused AI applications in this segment.

3. Vision Fine-Tuning

The introduction of vision fine-tuning has offered new room in the gpt-4o as it allows developers to fine-tune models not only by using text but also with images. This allows the model to learn and comprehend images more easily expanding the potential use cases such as better visual searches, object detection in self-driving cars, and more accurate interpretations of medical scans among others.

4. * Realtime API *

With the Realtime API, the task of developing speech-to-speech applications has become more seamless and does not require many tools. In the past such apps would have taken several steps: converting audio to text, processing the text with a language model, and text to speech. Now, OpenAI has an API that does all of this in an instant.

Function calling can be done via this API allowing applications to perform physical actions such as ordering goods or booking an appointment. The costs are $5 per million tokens of text placed into the API and $100 per million tokens of audio placed into the API proving an economy way of handling things in real time.

These updates make OpenAI one of the developers who can be swayed by the developer’s community who provide affordable and impressed tools. Cost reduction through prompt caching or enhanced capabilities through vision fine-tuning, there’s unprecedented flexibility and control for developers through OpenAI’s new features.

To understand how to use such possibilities, refer to the API section of OpenAI’s website.

Sources: OpenAI Blog, Developer Day Announcements

About the Author:

Amir Ghaffary – CEO of CLOXMEDIA – is on a relentless mission to revolutionize our grasp of the future, blending visionary insight with cutting-edge technology to craft a new paradigm of modern understanding. His work transcends traditional boundaries, bridging the gap between what is and what could be, inspiring a generation to rethink the possibilities of tomorrow. By advocating for a deeper integration of AI, digital transformation, and forward-thinking innovation, Amir is not just predicting the future—he’s actively shaping it, pushing society to embrace a bold new reality where technology and human potential are intertwined like never before.