OpenAI has presented a more affordable and slimmed-down version of its flagship artificial intelligence model.
The new product of the mentioned company, which is currently one of the largest players in the global area of machine intelligence and the developer of one of the most popular chatbots ChatGPT, aims to gain the attention of a wider range of vendors and business customers. It’s worth noting that the relevant actions of OpenAI are a reflection of the increasing competition that is currently being implemented in the worldwide market of artificial intelligence services. Against this background, companies are starting to come up with new solutions and develop additional versions of their products to preserve and sometimes increase the number of users in a favorable scenario.
On Thursday, July 18, OpenAI announced the GPT-4o mini. This virtual product is a smaller and cheaper modification of GPT-4o, the most powerful artificial intelligence model among all configurations of the advanced technology developed by the company.
OpenAI reported that the new version of its flagship product will be available to free users and paying subscribers of ChatGPT Plus and Team. It is worth noting that this category of consumers can already use the GPT-4o mini. A new digital product will be offered to enterprise customers next week.
The GPT-4o mini replaces the GPT-3.5 Turbo, an older artificial intelligence model, in ChatGPT.
OpenAI presented the GPT-4o in May. The company focused attention on the ability of its flagship artificial intelligence model to process audio and visual information in real-time. At the same time, the launch of some features of the powerful digital assistant has been postponed. The Company has made the appropriate decision due to the need to resolve certain security issues.
It is worth noting that during the period of active dissemination and use of artificial intelligence, the relevance of cybersecurity is increasing. In this context, it is important to understand that scammers also have access to advanced technology, which makes their activities more difficult to detect. Digital literacy is one of their effective tools for countering crime in the virtual space. For example, a query in an Internet search system, such as how to know if my camera is hacked, will allow anyone to get information about what signs and signals can detect unauthorized third-party interference in the functioning of a personal device.
OpenAI plans to improve the GPT-4o mini over time. In this case, it means expanding the list of functions of a smaller version of the company’s flagship artificial intelligence model.
It is worth noting that modifications of powerful AI systems, characterized by a decrease in the effectiveness of certain indicators, are gradually becoming a common product in the digital intelligence industry. For example, such players in the mentioned industry as Anthropic and Google are already involved in this practice. More flexible, less capable, and more financially affordable versions of powerful artificial intelligence models are in demand among developers. Such modifications of flagship AI systems are better suited for automating basic tasks of a large volume. At the same time, a larger artificial intelligence model is capable of doing more complicated work.
Olivier Godement, the head of product for OpenAI’s API, says that the company, as part of its mission to enable the bleeding edge, to build the most powerful, useful applications, wants to continue doing the frontier models, pushing the envelope here. It was also noted that the ChatGPT developer strives to have the best small models of artificial intelligence in the list of its products.
Olivier Godement stated that last week, some developers tested the GPT-4o mini. According to him, the e-mail startup Superhuman used this digital product to generate automatic replies to messages. He also stated that the fintech startup Ramp applied GPT-4o mini to pull information from receipts.
The GPT-4o mini can take in and generate text and images. OpenAI intends to expand the boundaries of the capabilities of this virtual product over time by introducing content processing functions of other categories. The company also announced that the GPT-4o mini is its first artificial intelligence model that uses a new safety tactic called Instruction Hierarchy. The corresponding approach provides that the AI system should give priority to certain instructions so that it is more difficult for users to force the digital assistant to perform actions related to the category of prohibited scenarios.
