The text-generating AI models GPT-3.5-turbo and GPT-4 have been updated by OpenAI to include a feature called function calling. Developers can now specify functions for GPT-4-0613 and GPT-3.5-turbo-0613, and the model will intelligently decide which JSON object to output that has the arguments needed to invoke those functions.
This is a novel approach to more securely link GPT’s capabilities with external tools and APIs, according to a blog post published by OpenAI on Tuesday. Developers can build chatbots that respond to queries by calling external tools (such as ChatGPT Plugins) using the function calling feature.
For instance, it can convert queries such as “What’s the weather like in Boston?” to get_current_weather(location: string, unit: ‘celsius’ – ‘fahrenheit’) or “Email Anya to see if she wants to get coffee next Friday” to a function call like send_email(to: string, body: string). OpenAI indicated that it will start the process of upgrading and deprecating the initial releases of GPT-4 and GPT-3.5-turbo, which it announced in March.
On June 27, all software utilizing the stable model names (GPT-3.5-turbo, GPT-4, and GPT-4-32k) will be automatically updated to the new models. Additionally, the business announced a 25% price reduction for the GPT-3.5-turbo.
Developers can use GPT-3.5-turbo for $0.0015 per 1,000 input tokens and $0.002 per 1,000 output tokens, yielding about 700 pages per dollar. A pricing reduction is also being applied to Text-embedding-ada-002, one of OpenAI’s most popular text embedding models.