OpenAI announces updates to API including improved function calling capabilities

As OpenAI has gained more popularity than ever thanks to the launch of ChatGPT last fall, its application programming interface (API) has also become an in-demand tool for developers. Various companies have found inspiration in OpenAI’s success and want to replicate it, with developers using its API to build more and more tools.

To help cope with this increase in demand, OpenAI has announced some changes to its API, such as improved function calling capabilities, a more steerable version of GPT-4, and GPT-3.5 Turbo, a new version of GPT-3.5 Turbo. 16k reference version, and the embedding model saw a 75% reduction in cost, resulting in lower costs for developers paying for the API.

Too: GitHub developer survey says 92% of programmers are using AI tools

Some of the new updates include a new function calling capability in the Chat Completion API that will allow developers to more reliably combine the power of the GPT model with external tools. With the update, developers will be able to provide instructions for GPT-4 and GPT-3.5 Turbo by describing functions, and the model will output a JSON object with the arguments needed to call those functions.

The update makes it easier for developers to create chatbots or applications that interact with external tools and APIs to perform specific tasks, such as sending email, getting weather or flight information, or extracting data from text sources such as websites.

Too: GPT-3.5 vs GPT-4: Is ChatGPT Plus Worth Its Subscription Fee?

The updates also make for more portable GPT-4 and GPT-3.5 models, so developers can have more control over the model’s output. OpenAI is allowing developers to create context, specify desired formatting, and provide instructions to the model about desired output. Essentially, developers have more say over the tone, style, and content of the responses generated by the models they use in their applications.

OpenAI also announced the launch of a new 16k reference version of GPT-3.5 Turbo, which differs from the GPT-3.5 model behind ChatGPT as it is specifically tailored for developers building chat-based applications. Was. The latest 16k model is an upgraded version from the standard, which the 4k model previously used.

Too: AMD introduces MI300x AI chip as ‘Generative AI accelerator’

Context in “16k context” is used to describe text within a conversation that helps provide context and helps the model understand the input and provide relevant feedback to the conversation. The 4k, or 4,000 tokens in the standard GPT-3.5 Turbo model, limit the model’s ability to retain context in a conversation to a few paragraphs. 16k, or 16,000, tokens equals about 20 pages of text, giving the model a large amount of text to reference.

Finally, OpenAI announced that it was able to become more efficient and reduce costs, so it reduced the embedding model by 75% to $0.0001 per 1k token, and GPT-3.5 Turbo by 25% to $0.0014 per 1k input token and $0.002 per 1k token. is outputting 1k. token. The new GPT-3.5 Turbo-16k model costs $0.003 per 1k input token and $0.004 per 1k output token.