Microsoft's OpenAI has undoubtedly been leading the AI space with its cutting-edge technological advancements and pioneered exploration of the technology's untapped potential. The company keeps bringing new, more enhanced versions of ChatGPT that are better equipped and have greater capabilities. The latest iteration was the GPT-4 turbo, which was said to upgrade performance and improve cost efficiency. The platform keeps evolving, and on Thursday, OpenAI announced a new small AI model that will outperform the current versions and is available for developers and consumers to access.
OpenAI announced a new small AI model, GPT-4o mini, that will exceed performance expectations and surpass current AI models of its size
OpenAI is focused heavily on making ChatGPT's experience comprehensive by introducing more enhanced capabilities with every version. To continue improving, the company rolled out a new version that will be lightweight and economical so users can get true value for money.
The latest small AI model is called the GPT-4o mini. It is said to offer more advanced performance and require less power, saving on resources and targeting a wider consumer base with its affordable cost. This new model would replace GPT-3.5 Turbo for the smallest AI model yet.
Developers wanting to use OpenAI models for building apps earlier incurred a higher cost, so the light models introduced by other companies, such as Google's Gemini 1.5 Flash, gained more popularity. By bringing a smaller, lighter model, OpenAI is stepping up its game, which could outperform all other AI models and be popular.
OpenAI's claim that the model's performance surpasses existing models is backed up by a comparison of different models drawn by Artificial Analysis. The GPT-4o mini scored 82 percent on Massive Multitask Language Understanding (MMLU), signifying the model's ability to understand different contexts and expand over wider domains. In comparison, Google's Gemini Flash scored 77.9 percent, and Anthropic's Claude Haiku scored 73.8 percent, significantly below the new model's multitasking abilities.
Since smaller AI models rely on less power, they tend to be more energy efficient and save on costs. Due to limited computational resources being used, they are designed to be more efficient and offer advanced performance. OpenAI, by developing its own compact and light model, could offer robust capabilities to perform complex tasks more effectively.
OpenAI's claim that GPT-4o might outperform leading smaller language models by being more cost-efficient, smarter, and faster does seem promising, given its plans to add audio and video support later on. ChatGPT Free, Plus, and Team users can start using the new model, and enterprise users will be given access next week.
Follow Wccftech on Google to get more of our news coverage in your feeds.
