Information Technology: Faculty Publications

A Comparative Analysis of OpenAI's API Performance: Insights from AI Response Statistics

Document Type

Conference Proceeding

Publication Date

7-15-2025

Publication Title

2025 6th International Conference on Artificial Intelligence, Robotics and Control (AIRC)

DOI

10.1109/AIRC64931.2025.11077529

Abstract

Generative AI models have the capability to produce a wide range of new content based on the data they are trained on. These models can generate not just text, but also various types of multimedia content, including images. In recent years, they have become increasingly popular due to their significant impact across multiple fields. They are used in various applications, from text and image generation to music creation, as well as in education, healthcare, robotics, finance, education, autonomous vehicles, and more. However, these models face numerous challenges, such as overfitting the outputs, inconsistency in the results, response time, and token usage. This study presents a comparative analysis of the performance of OpenAI's API, focusing on insights gleaned from AI response statistics. The evaluation encompasses various performance insights on different model approaches, such as response tokens, processing speed, and result token efficiency. Using quantitative metrics and qualitative assessments, the research identifies trends and patterns in API performance, highlighting strengths and areas for improvement. The study concludes with recommendations for optimizing API functions, contributing to the broader discourse on advancing AI integration in real-world applications.

Comments

Georgia Southern University faculty member, Atef Mohamed, Hayden Wimmer, and Christopher A. Kadlec co-authored, "A Comparative Analysis of OpenAI's API Performance: Insights from AI Response Statistics."

Link to Full Text

Share

COinS