Unlock the New Features of OpenAI API

Unlock the New Features of OpenAI API

Table of Contents:

1. Introduction
2. Overview of the OpenAI API update
3. Convenience and cost reduction 3.1. Token limit relaxation 3.2. Lower cost of using the API
4. Function Calling feature 4.1. Introduction to Function Calling 4.2. Models associated with Function Calling 4.3. How to use the Function Call API
5. Example of using the Function Call API
6. Conclusion

Article

Introduction

Greetings everyone, it's AIVTuber Nyanta here. Today, I'll be sharing some exciting news about the latest OpenAI API update. OpenAI recently released new functions as part of their API which have garnered a lot of Attention. In this article, I'll provide an in-depth explanation of one of these new features known as Function Calling. I'll also discuss the convenience and cost benefits offered by this API update.

Overview of the OpenAI API update

On June 13, OpenAI introduced several new functions through an API update. These functions bring enhanced usability and cost-effectiveness to the OpenAI platform. To fully understand the capabilities and advantages of Function Calling, it's important to explore the broader Context of this API update.

Convenience and cost reduction

Token limit relaxation

One significant improvement in the API is the relaxation of the token limit. Previously restricted to just over 4,000 tokens, the introduction of a new model called gpt-3.5-turbo-16k allows handling up to 16,000 tokens. This update provides users with more freedom to generate longer and more detailed outputs. The increased token limit opens up new possibilities for creative applications.

Lower cost of using the API

Another noteworthy aspect of the OpenAI API update is the reduced cost associated with using the API. The introduction of a feature called Embeddings provides vectorization of documents for document search purposes at a 75% discount. Additionally, the cost of input tokens has been reduced by 25% to $0.0015 per 1000 tokens. The output cost has also been reduced to $0.002 per 1000 tokens. These cost reductions make the API more affordable and accessible to a wider range of users.

Function Calling feature

Introduction to Function Calling

The most exciting addition to the OpenAI API update is the Function Calling feature. Function Calling can be considered as a plug-in function for the Web version. It outputs third-party APIs and other Relevant information in a format that is easily callable. By integrating Function Calling, OpenAI offers users a more streamlined and convenient way to invoke external functionalities.

Models associated with Function Calling

Two models, gpt-4-0613 and gpt-3.5-turbo-0613, are central to the Function Calling feature. These models receive input, determine the appropriate function to call, and output the necessary information in a JSON format compatible with the input requirements of the target function. Think of this process as employing a rung-chain agent that transforms the output into a usable format to execute the desired function.

How to use the Function Call API

To use the Function Call API, You'll first need to set up an APIKey provided by OpenAI. Next, define the function you want to call, specifying the parameters it requires and the intended output. An example function called "getCurrentWeather" is provided, which retrieves weather information Based on location and units. If you have multiple functions, ensure to describe each one thoroughly.

Once you've defined the function, you can execute the model. Specify "gpt-3.5-turbo-0613" as the model argument and provide the user's message. Additionally, indicate the specific function using FunctionCall="auto". This setup enables the system to determine when to utilize the external function to generate the desired output.

Example of using the Function Call API

Let's consider an example to illustrate how the Function Call API works. Suppose we ask the language model, "What is the weather like in Boston?" The Function Call API will determine whether the function "getCurrentWeather" is required to answer this question. If so, it will extract the location and unit information from the question and present them as inputs required by the "getCurrentWeather" function.

For instance, the Function Call API recognizes the need for the "getCurrentWeather" function and determines that the location is Boston and the unit is Celsius. It then outputs these variables, enabling the execution of the "getCurrentWeather" function externally. The output can be conveniently obtained in JSON format and incorporated into further language models to generate comprehensive answers to user queries.

Conclusion

The Function Calling feature and other API updates introduced by OpenAI offer enhanced convenience, flexibility, and cost effectiveness. The relaxation of token limits, reductions in API cost, and the functionality provided by Function Calling make the OpenAI API more accessible and practical. By simplifying the integration of third-party APIs, OpenAI empowers users to harness the full potential of the language models and elevate their applications to new heights.

Highlights

• OpenAI API has been updated, introducing new features like Function Calling.
• Users can now enjoy added convenience, lower costs, and an increased token limit.
• Function Calling serves as a plug-in for invoking external functionalities easily.
• Two models, gpt-4-0613 and gpt-3.5-turbo-0613, are central to Function Calling.
• The Function Call API enables users to generate outputs in a format compatible with external functions.
• Examples showcase how Function Calling can be utilized to achieve desired results.

FAQ

Q: What are the advantages of the OpenAI API update? A: The OpenAI API update brings several advantages, including increased convenience, reduced costs, and a higher token limit.

Q: How does the Function Calling feature work? A: Function Calling allows users to invoke external functions easily by converting outputs into a format compatible with those functions.

Q: Which models are associated with Function Calling? A: gpt-4-0613 and gpt-3.5-turbo-0613 are the models specifically designed for Function Calling.

Q: Can you provide an example of using the Function Call API? A: Certainly! By using the "getCurrentWeather" function as an example, the Function Call API can retrieve weather information based on location and units.

Q: How can the Function Call API be used effectively? A: Defining the functions, specifying inputs and outputs, and calling the relevant functions appropriately can ensure effective utilization of the Function Call API.