OpenAI Introduces Economical GPT-4o Mini AI Model

OpenAI has launched GPT-4o Mini, a new AI model accessible to more people and costing less. There is a lot of competition in the field of AI these days, especially from big names like Google and Meta.

What is the GPT-4o Mini?

GPT-4o Mini is a smaller, less expensive version of OpenAI’s other language models. It costs 15 cents per million tokens in and 60 cents per million tokens out, more than 60% less than its predecessor, GPT-3.5 Turbo.

Performance metrics

OpenAI says GPT-4o outperforms GPT-4 in the mini chat version and scores 82% on the MMLU benchmark for massively multitasking language understanding. MMLU tests how well a model can learn and use the language in different situations. On the other hand, Google’s Gemini Flash scored 77.9% and Anthropic’s Cloud Haiku scored 73.8%. This makes GPT-4o Mini very competitive in terms of its ability to understand words.

Availability

Mini forms require less computing power, so they can work for companies that don’t have many resources. Because it is so efficient, more companies can use creative AI in their work. ChatGPT users at Free, Plus and Team levels can immediately access GPT-4o Mini instead of GPT-3.5 Turbo. Next week, business users will be able to do the same. The model is based on what we know between now and October 2023.

More about OpenAI

OpenAI was started as a non-profit group in December 2015. OpenAI Gym, a set of tools for creating reinforcement learning algorithms, was its first major release. In 2019, OpenAI’s GPT-2 model caused a stir because it could generate text that was not as it appeared. In 2020, OpenAI created rules to properly use its AI systems. The company’s name shows that it wants to encourage open collaboration in AI research, but it turned to a capped-profit model to raise the money needed to fund its work.

Source

Cave on the Moon: advances in space exploration