OpenAI has unveiled GPT-4, the latest iteration of its widely used artificial intelligence chatbot, ChatGPT. The new model is capable of processing up to 25,000 words, roughly eight times more than its predecessor.

Additionally, it can respond to images, writing captions and descriptions, and providing recipe suggestions from photographs of ingredients.

ChatGPT has been popular for its ability to mimic human-like language and can generate various writing styles such as songs, poems, and marketing copies.

However, educators caution against its use in academic settings. OpenAI has spent six months developing safety features for GPT-4, but warns of its potential to propagate disinformation.

The service will initially be available to ChatGPT Plus subscribers, with Microsoft’s Bing search engine already using it.

According to OpenAI, GPT-4 possesses “enhanced reasoning abilities” in comparison to ChatGPT. The model can perform tasks such as finding appropriate meeting times for three schedules.

Additionally, OpenAI has unveiled partnerships with language-learning platform Duolingo and Be My Eyes, an app designed to assist the visually impaired, to develop AI chatbots capable of assisting users using natural language.

Despite its advanced capabilities, OpenAI has cautioned that GPT-4 is not completely dependable and may “hallucinate,” meaning it could invent information or make logical mistakes, similar to its predecessors.