Sam Altman, CEO of OpenAI, unveils new and improved GPT-4 Turbo

A number of improvements to OpenAI's large language models, GPT-4 and GPT-3.5, have been revealed. These enhancements include a substantially larger context window and updated knowledge bases. OpenAI pledges to protect its clients from copyright litigation, following in the footsteps of Microsoft and Google.

GPT-4 Turbo was unveiled by OpenAI and is presently available through an API preview. It has been trained with data up to April 2023. Compared to the previous GPT-4 version, which was only updated to March 2021, this represents a major advancement. Though a precise release date has not been revealed, OpenAI plans to release a production-ready Turbo model in the upcoming weeks.

A larger view is provided by GPT-4 Turbo, which has a 128K context window—that is, more than 300 pages of text in a single prompt. Greater context windows allow GPT to understand questions more thoroughly and offer more intelligent answers. The context windows used in previous GPT-4 versions were 8K and 32K.

Additionally, OpenAI emphasised GPT-4 Turbo's affordability for developers. Compared to GPT-4, where input costs USD 0.03, this model will only cost USD 0.01 per 1,000 tokens, which is the basic unit of text or code that large language models must process. The cost of each output is USD 0.03 for every 1,000 tokens. In conclusion, OpenAI claims that the new GPT-4 is three times more economical than the previous iteration.

According to OpenAI, GPT-4 Turbo's improvements enable users to give the model instructions to perform increasingly complicated tasks with just one prompt. Users also have the option to designate a specific coding language, like JSON or XML, for the results.

Businesses that use the GPT-3.5 Turbo model will also gain from enhanced functionality, pricing, and context windows. This version includes the same functional updates as GPT-4 Turbo and comes with a default 16K context window. The cost of the GPT-3.5 Turbo is USD 0.01 for input and USD 0.002 for output.

GPT-3.5 Turbo was first released by OpenAI in March, and it was positioned as the best model for non-chat applications. In August, OpenAI released a version that can be fine-tuned. 

Apart from enhancing the model, OpenAI has declared that it will offer enterprise users copyright indemnity, following the practises of Google and Microsoft. This programme, called Copyright Shield, represents OpenAI's promise to defend clients against lawsuits alleging copyright infringement involving generally accessible ChatGPT Enterprise features and OpenAI's developer platform. The business released a statement saying, "We will now step in and defend our customers, and pay the costs incurred, if you face legal claims around copyright infringement."

Also Read: Zomato CEO Deepinder Goyal Reveals the Truth About Zomato Discounts: “They are not very big, they only appear so”