OpenAI, the maker of ChatGPT, said Thursday it is launching the GPT-4o mini, a low-cost miniature AI model that aims to make its technology more affordable and less energy-intensive, allowing startups to use allows targeting a wider pool of
Microsoft-backed ( MSFT.O ) opens new tab OpenAI, the market leader in the AI software space, is working to make it cheaper and faster for developers to build applications based on their models, as When deep rivals like Meta ( META.O ) open a new tab and Google ( GOOGL.O ), open a new tab rush to gain a bigger market share.
At 15 cents per million input tokens and 60 cents per million output tokens, GPT-4o Mini is more than 60 percent cheaper than GPT-3.5 Turbo, OpenAI said.
OpenAI said it currently outperforms the GPT-4 model on chat preferences and scores 82% on Massive Multitask Language Understanding (MMLU).
MMLU is a text intelligence and reasoning benchmark used to test the capabilities of language models. A high MMLU score indicates that one can better understand and use language in a variety of domains, enhancing real-world use.
According to OpenAI, the GPT-4o mini model scores 77.9% for Google’s Gemini Flash and 73.8% for Anthropic’s Cloud Haiku.
Smaller language models require less computational power, making them a more affordable option for companies with limited resources looking to deploy generative AI in their operations.
With MiniModel currently supporting text and vision in the application programming interface, OpenAI said support for text, image, video and audio input and output will be made available in the future.
OpenAI said that ChatGPT’s free, Plus and Team users will be able to access GPT-4o mini – which is expected to arrive in October 2023 – from Thursday, replacing GPT-3.5 Turbo, for enterprise users. Access will be available from next week.
