OpenAI's GPT-4 Turbo in Preview: Offering Developers a 66% Cost Reduction and Enhanced Copyright Protection for Customers.
OpenAI is taking strides to enhance its substantial language models, GPT-4 and GPT-3.5, introducing upgraded knowledge bases and significantly expanding the context window. In a noteworthy move following the footsteps of industry giants like Google and Microsoft, OpenAI is now committed to safeguarding its customers against potential copyright disputes.
At their inaugural developer conference, OpenAI made a significant announcement regarding GPT-4 Turbo, currently accessible through an API preview. This advanced iteration has been meticulously trained on information extending all the way up to April 2023, marking a substantial leap from the previous GPT-4 version released in March, which was only informed by data up to September 2021. OpenAI is gearing up to unleash a production-ready Turbo model in the coming weeks, although the exact release date remains undisclosed.
The latest iteration of GPT-4 Turbo introduces a significant expansion of its data comprehension capabilities, boasting an impressive 128K context window, which OpenAI equates to over 300 pages of text in a single prompt. This larger context window empowers large language models like GPT-4 to grasp more nuanced questions and deliver more comprehensive and thoughtful responses. Notably, OpenAI had previously released two versions of GPT-4, one with a mere 8K context window and another with a 32K context window.
OpenAI has made strides in cost-efficiency, making GPT-4 Turbo more developer-friendly. Input costs have been slashed to just $0.01 per 1,000 tokens, compared to the $0.03 on GPT-4, with output priced at $0.03 per 1,000 tokens. As a result, OpenAI claims that the latest GPT-4 version is three times more economical than its predecessors.
Furthermore, GPT-4 Turbo retains its versatility by accepting image prompts, text-to-speech requests, and integrating DALL-E 3, a feature initially introduced in October. These enhancements empower users to pose more complex questions to the model, allowing them to specify their desired coding language, such as XML or JSON, for the results.
Not to be outdone, companies utilizing the GPT-3.5 Turbo model can also anticipate improvements in context windows, functionality, and pricing. This version will offer a default 16K context window and benefit from the same feature upgrades as GPT-4 Turbo. GPT-3.5 Turbo comes at a cost of $0.01 for input and a mere $0.002 for output.
OpenAI, known for its forward-thinking approach, extends its commitment to innovation by following in the footsteps of tech giants like Microsoft and Google. The company introduces "Copyright Shield," a program aimed at providing copyright protection to enterprise users, further enhancing its value proposition.
In a blod moves to stand by their users, OpenAI has announced their commitment to protect and support their customers when faced with legal challenges related to copyright infringement. OpenAI's pledge encompasses the safeguarding of commonly accessible features within ChatGPT Enterprise and the OpenAI developer platform.
This initiative mirrors the steps taken by tech giants Google and Microsoft, both of which have offered similar protective measures to their customers in cases of copyright infringement claims stemming from the use of their embedded generative AI features and Copilot AI products, respectively.
