OpenAI Previews GPT-4 Turbo Prior to Upcoming Reduced Price Production Release

AI research and deployment company OpenAI launched a preview of their next generation AI model, GPT-4 Turbo.  GPT-4 is a large multimodal AI model that can accept text or image inputs and output text for use in solving difficult problems.  OpenAI released the first version of GPT-4 in March 2023 and made GPT-4 generally available to all developers in July. GPT-4 is optimized for chat but can also be used for traditional completions tasks using the Chat Completions API.  

The next generation model, GPT-4 Turbo, is more capable and has knowledge of world events up to April 2023. GPT-4 Turbo has a 128k context window, so it can fit the equivalent of more than 300 pages of text in a single prompt. The model’s performance has also been optimized, so that GPT-4 Turbo input tokens are now three times less expensive and output tokens two times less expensive compared to GPT-4.

Function Calling Updates

The new Turbo release features several functional improvements, including the ability to send one message requesting multiple actions, such as “open the car window and turn off the A/C”, which would previously require multiple roundtrips.  The new release also features improved function calling accuracy, with GPT-4 Turbo more likely to return the right function parameters.

Reproducible Outputs and Log Probabilities

The new seed parameter enables reproducible outputs by making the model return consistent completions most of the time. This beta feature is useful for use cases, such as replaying requests for debugging, writing more comprehensive unit tests, and generally having a higher degree of control over the model behavior. OpenAI has been using this feature internally for their own unit tests and found it invaluable. 

They are also launching a feature to return the log probabilities for the most likely output tokens generated by GPT-4 Turbo and GPT-3.5 Turbo in the next few weeks, which will be useful for building features, such as autocomplete in a search experience.

GPT-4 Turbo is currently available for all paying developers with plans to release the stable production-ready model in the coming weeks.

Updated GPT-3.5 Turbo

In addition to GPT-4 Turbo, OpenAI released a new version of GPT-3.5 Turbo that supports a 16K context window by default as well as improved instruction following, JSON mode, and parallel function calling. Applications using the gpt-3.5-turbo name will automatically be upgraded to the new model on December 11. Older models will continue to be accessible until June 13, 2024.