OpenAI launches GPT-4o mini, a cheaper version of its AI

OpenAI announced that the GPT-4o mini will replace the GPT-3.5 Turbo as the company’s smallest model, although its main drawback is that it will not have the multimodal capabilities of the more advanced models.

However, this new option is intended to offer alternatives to users who are recently joining the generative Artificial Intelligence fever, based on products that are less complex to understand.

“To empower every corner of the world with AI, we need to make models much more affordable,” OpenAI’s head of product APIs, Oliver Godement, told various US media outlets.

This new tool will be priced at 15 cents per million input tokens and 60 cents per million output tokens. The company also reported that its context window is approximately one book and uses information with a cutoff date until October 2023.

Compared to other models in its class, OpenAI said GPT-4o mini is on par with Meta’s Llama 3 8b or Google’s Gemini 1.5 Flash. However, the company said it is more cost-effective because of its speed.

Some startups have already used OpenAI’s new technology. Financial startup Ramp, for example, used it to build an app that extracts expense data from receipts and sorts it.

Godement said that this new model “is going to be very popular” due to the potential it will offer for the creation of simple applications for both expert and inexperienced developers.