A new AI model, GPT-4o mini, has been launched by OpenAI, the leading company in the AI software space backed by Microsoft. The release on Thursday introduced this cost-efficient and energy-saving model, designed to make advanced AI technology more affordable and accessible, targeting a broader customer base.
The GPT-4o mini has been priced at 15 cents per million input tokens and 60 cents per million output tokens, marking it over 60% cheaper than GPT-3.5 Turbo. Despite the lower cost, the GPT-4o mini has been shown to outperform the larger GPT-4 model in chat preferences and achieved an impressive 82% on the Massive Multitask Language Understanding (MMLU) benchmark. This score surpasses Google’s Gemini Flash at 77.9% and Anthropic’s Claude Haiku at 73.8%.
MMLU is a benchmark used to assess the capabilities of language models in understanding and using language across various domains, indicating the model’s real-world applicability. The high MMLU score of GPT-4o mini signifies its superior textual intelligence and reasoning abilities.
The commitment of OpenAI to making AI technology more accessible has been demonstrated with the introduction of GPT-4o mini. Smaller language models like GPT-4o mini require less computational power, offering a more affordable option for businesses looking to integrate generative AI into their operations.
Currently, support for text and vision inputs through its application programming interface (API) has been provided by the GPT-4o mini. Plans to expand its capabilities to include text, image, video, and audio inputs and outputs have been announced. Access to GPT-4o mini will be available starting Thursday for ChatGPT Free, Plus, and Team users, replacing GPT-3.5 Turbo. Enterprise users will gain access next week.
The broader strategy of OpenAI to lead in “multimodality” has been highlighted by the launch of GPT-4o mini. This strategy integrates various types of AI-generated media such as text, images, audio, and video into a single tool, reflecting the multifaceted way humans interact with the world.
Founded in 2015 and now valued at over $80 billion, OpenAI faces the challenge of staying ahead in the competitive generative AI market while managing the substantial costs associated with developing and maintaining its models. The introduction of the GPT-4o mini is seen as a strategic move to balance these demands by offering a powerful yet cost-effective solution.
The importance of multimodality was emphasized by Chief Operating Officer Brad Lightcap in a previous statement: “The world is multimodal. If you think about the way we as humans process the world and engage with it, we see things, we hear things, we say things — the world is much bigger than text. So to us, it always felt incomplete for text and code to be the single modalities, the single interfaces that we could have to how powerful these models are and what they can do.”