.png)
OpenAI has recently unveiled GPT-4 Turbo, an enhanced iteration of its AI model GPT-4, during its developer conference. The model is currently in preview mode for paid developers through an API, and is set to be more widely accessible in the upcoming weeks.
GPT-4 Turbo comes in two variants: one that focuses on text analysis and another that delves into text and image context comprehension. Notably, this model is equipped to retrieve information up until April 2023, expanding the horizon beyond the limitations of its predecessor which only had knowledge until September 2021. Subsequently, GPT-4 was upgraded to possess information until January 1, 2022, and now extends its capacity to April 2023. The enhanced GPT-4 Turbo is designed to tackle more intricate tasks based on individual prompts.
GPT-4 Turbo boasts a 128K context window, enhancing its ability to grasp the nuances of queries and generate more nuanced responses. To provide a frame of reference, the previous GPT-4 model offered versions with 8K and 32K context windows.
According to OpenAI, developers can now access GPT-4 Turbo at a reduced cost of $0.01 per 1000 tokens, compared to the $0.03 per 1000 tokens required for GPT-4. This new pricing structure makes GPT-4 Turbo three times more cost-effective than its predecessor. The pricing of GPT-4 Turbo varies based on image size, with a 1080x1080 pixel image costing approximately $0.00765 as per OpenAI's pricing details. Apart from introducing GPT-4 Turbo, OpenAI also unveiled GPTs during its developer conference. These customizable versions of ChatGPT allow users to create tailored assistants such as a 'writing coach', 'sous chef', or 'math teacher' without the need for any programming knowledge. Additionally, OpenAI plans to launch a GPT Store later in the month, providing an avenue for creators of GPTs to monetize their creations.
English (US)