OpenAI Major Update for ChatGPT: Function call & Improved

On June 13, OpenAI suddenly released a major update for ChatGPT. The main updates include:

Added function call feature in Chat Completions API
Improved GPT-4 and GPT-3.5-turbo versions
Updated the 16k context version of GPT-3.5-turbo
Reduced the cost of embedded models by 75%
Reduced the input tokens cost of GPT-3.5-turbo by 25%
Announced the deprecation schedule for GPT-3.5-turbo-0301 and GPT-4-0314 models

OpenAI also announced that they would invite more people to try GPT-4 in the coming weeks until the waitlist is entirely removed.

Content Guideline

What is a function call?

Developers can now describe functions to the GPT model, allowing the model to intelligently choose to output a JSON object containing parameters to call these functions.

In simpler terms, GPT will know when it needs to use external tools to solve a problem. It eliminates the need for developers to describe complex prompts and further reduces the likelihood of errors.

Example

When you ask ChatGPT, “What’s the weather like in Boston?” Your question will be interpreted by the model as calling a third-party weather API. It will retrieve relevant data and output an accurate answer.

OpenAI provides a standardized interface for calling, giving developers greater flexibility when developing GPT-based applications.

New models

OpenAI has released the improved models GPT-4-0613 and gpt-3.5-turbo-0613. Both support the function call feature.

OpenAI has also released the gpt-3.5-turbo-16k model, supporting a context length of 16,000 tokens. Four times larger than the current GPT-3.5-turbo.

ChatGPT’s inability to input long text has always been a criticized issue. Now, with the gpt-3.5-turbo-16k model, you can input up to 20 pages of text. That allows ChatGPT to read long documents and papers directly.

The gpt-3.5-turbo-16k provides more value without extra cost. Although the text volume is four times larger, the price is only twice that of GPT-3.5-turbo.

OpenAI also announced that they would further reduce the cost of running models. For example, the input tokens cost of GPT-3.5-turbo has been reduced by 25%. That means you can process approximately 700 pages of documents per dollar. For more details check here!

OpenAI Ongoing Efforts Summary

Following the major update to ChatGPT, OpenAI continues to focus on enhancing its AI technology and making it more accessible to developers and users. Here’s a summary of their ongoing efforts:

Continuous improvement

OpenAI is committed to refining and expanding the capabilities of its models. They actively seek user feedback to identify areas for improvement. It ensures that their AI technology remains cutting-edge and relevant to users’ needs.

Open access

By gradually inviting more people to try GPT-4 and eventually removing the waitlist. OpenAI aims to make its AI technology more widely available. This will help drive innovation and allow more developers to benefit from the advanced features of GPT-4.

Extensive documentation

OpenAI provides extensive documentation and resources to help developers better understand and utilize the features offered by their AI models. This includes guides on how to use function calls, implement new models, and optimize API usage.

Flexible pricing

With the recent reduction in embedded model costs and input token costs. OpenAI demonstrates its commitment to making AI technology more affordable for developers. This allows for greater experimentation and more widespread adoption of AI-powered solutions.

Community support

OpenAI fosters a strong community of developers, researchers, and users who share their experiences, insights, and best practices. This collaborative environment helps drive innovation.

What Should OpenAI’s ChatGPT Consider

There are a few more aspects worth considering when exploring OpenAI’s ChatGPT and its updates:

Privacy and security

OpenAI is dedicated to maintaining user privacy and ensuring data security. They continually update their data handling practices and work to comply with industry standards and regulations.

New use cases

As OpenAI improves its models and adds new features, developers can explore novel applications for AI in various industries. Such as healthcare, finance, marketing, education, and more. The versatility of ChatGPT enables it to cater to a wide range of needs and requirements.

Multilingual support

While the current focus is primarily on English language models, OpenAI has shown interest in expanding its AI capabilities to handle other languages. This would make the technology even more accessible and useful for a global audience.

Performance monitoring

OpenAI is committed to tracking the performance of its models and addressing any limitations or biases that may arise. By actively monitoring and refining the models, OpenAI ensures that the AI technology remains accurate, reliable, and useful for various applications.