GPT-4 Turbo is OpenAI’s most powerful large language model yet

OpenAI introduced GPT-4 Turbo at its first developer conference, a new iteration of its language model that accepts inputs up to 300 pages long, significantly surpassing the 50-page limit of its predecessor. This enhancement allows for more complex prompts and potentially more insightful responses. The model has been updated with data up to April 2023, a substantial leap from the previous cutoff in September 2021, although the earlier GPT-4 model could access the internet for current information.

GPT-4 Turbo now also supports image prompts, offering the ability to generate captions or descriptions, and can handle text-to-speech requests. Additionally, users can upload documents for analysis, a feature already present in some competing chatbots. The cost for developers to use this new model will be three times less expensive than before, as OpenAI has reduced the price for tokens required for processing instructions and generating responses.

Alongside the new model, OpenAI announced that ChatGPT has reached over 100 million weekly active users globally, with adoption by more than 92% of Fortune 500 companies. OpenAI also committed to legally and financially protecting its customers, including enterprises, from copyright infringement claims related to the use of its products. Furthermore, OpenAI presented the concept of “mini-ChatGPTs,” which are specialized tools for single tasks that can be created without coding knowledge and shared or sold in an OpenAI store. The release date for GPT-4 Turbo to the general public remains unannounced, with current access to GPT-4 available at $20 per month.