OpenAI Releases GPT-4 API, Announces Retirement of Older Models
Starting today, all paying API customers can access the GPT-4 API. OpenAI plans to expand access to new developers by the end of the month and gradually raise rate-limits based on compute availability.
To meet production-scale use requirements, OpenAI has made the GPT-3.5 Turbo, DALL·E, and Whisper APIs generally available as well. Fine-tuning capabilities for GPT-4 and GPT-3.5 Turbo are currently in progress and expected to be available later this year.
Since its introduction in March, the Chat Completions API has dominated OpenAI’s GPT usage, accounting for 97% of API utilization.
The earlier Completions API, launched in June 2020, offered freeform text prompts for interacting with language models. However, OpenAI has discovered that better results can often be achieved with a more structured prompt interface.
The Chat Completions API’s structured interface, including system messages and function calling, along with multi-turn conversation capabilities, empowers developers to build conversational experiences and diverse completion tasks.
Furthermore, it enhances security by separating user-provided content from instructions, minimizing the risk of prompt injection attacks.
As OpenAI continues to invest in the Chat Completions API and optimize compute capacity, older models utilizing the Completions API will be retired in six months. While the API will remain accessible, OpenAI will designate it as “legacy” in the developer documentation starting today.
Effective from January 4, 2024, the older completion models will no longer be available.