On July 18th, the startup OpenAI announced the launch of a new artificial intelligence model called GPT-4o mini, which is more affordable and consumes less energy, aiming to expand its customer base.
ChatGPT's parent company stated that the GTP-4o mini is 60% cheaper than the GPT-3.5 Turbo model launched in 2023, with costs of just 15 cents per million input tokens and 60 cents per million output tokens.
Furthermore, this new model surpasses GPT-4 in chat preferences and scored 82% in the Multitasking Language Understanding Large (MMLU) test, higher than Google's Gemini Flash (77.9%) and Anthropic's Claude Haiku (73.8%).
MMLU is a standard for reasoning and textual intelligence used to assess the ability of language models. A higher MMLU score indicates that the model can understand and use the language better across a wider range of domains, enhancing its real -world usability.
Smaller language models require less computing power, making them a more affordable option for companies with limited resources looking to implement generative AI applications in their operations.
According to OpenAI, the current version of GPT-4o mini only supports text and image functionality in its application programming interface. Support for input and output of text, images, video , and audio will be integrated into the model at a later stage.
The American startup also stated that users of ChatGPT's Free, Plus, and Team plans will be granted access to GPT-4o mini instead of GPT-3.5 Turbo starting July 18th, while enterprise users will have to wait until next week to gain access.
Source: https://nhandan.vn/openai-ra-mat-mo-hinh-ai-moi-gpt-4o-mini-post819898.html
Comment (0)