OpenAI Unveils Cheaper, More Powerful GPT-4 Turbo Model
OpenAI introduces GPT-4 Turbo at its first developer conference, offering a larger context window and reduced pricing. The upgraded model features an updated knowledge base and the ability to process both text and images.
OpenAI unveils GPT-4 Turbo at its first-ever developer conference, presenting an upgraded version of its flagship AI model that is both more powerful and less expensive. The new model comes in two versions, with a strictly text-analyzing option available in preview via an API starting today and a second version that understands both text and images arriving in the coming weeks. OpenAI optimizes the performance to offer GPT-4 Turbo at a price that is three times cheaper for input tokens and two times cheaper for output tokens compared to the standard GPT-4.
The upgraded model boasts a significantly expanded context window of 128,000 tokens, which is four times the size of the previous version and the largest available in any commercial model. This larger context window allows the AI to consider much more text before generating additional text, reducing the likelihood that it "forgets" earlier parts of a conversation and veers off topic. The text-analyzing version costs $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens, while image-processing pricing varies depending on image size.
GPT-4 Turbo also features a more recent knowledge base to draw from when responding to user requests. While the original GPT-4 is trained on web data up to September 2021, the new model extends this knowledge cut-off to April 2023. This update means that the model yields more accurate answers to questions about recent events that occurred prior to the new cut-off date, making it a much more reliable tool for current information.