• OpenAI has announced new version for two of its generative AI offerings – GPT-3.5-turbo and GPT-4.
• Both models now feature a new capability called function calling.
• The new feature allows devs to describe programming functions and have the models create code to execute those functions.

OpenAI, the company behind the wildly popular ChatGPT generative AI platform, has announced new versions for two of its models.

The company explained that a new capability has been added to GPT-3.5-turbo and GPT-4, with the latter being the latest iteration of the platform that has been officially available to the public.

As for the new feature, it is called function calling. The company explained that the new feature will power the the Chat Completions API. To that end, it allows developers to describe programming functions to either of the models, and have said models create code to execute those functions.

“This is a new way to more reliably connect GPT’s capabilities with external tools and APIs,” the company highlighted in a blog post.

“These models have been fine-tuned to both detect when a function needs to be called (depending on the user’s input) and to respond with JSON that adheres to the function signature. Function calling allows developers to more reliably get structured data back from the model,” it added.

Some examples of the elements these models can generate include, chatbots that answer questions by calling external tools, as well as converting natural language into API calls or database queries.

Along with the new versions, OpenAI announced some pricing reductions on a pair of aspects to its models. Here, a 75 percent cost reduction on its embeddings model, along with a 25 percent cost reduction on input tokens for GPT-3.5-turbo.

It therefore looks like OpenAI is very much focused on incrementally improving its existing models, along with making them available on more platforms, such as a dedicated iOS app in recent weeks.

[Image – Photo by Mariia Shalabaieva on Unsplash]