OpenAI announced that all paying API customers will now have access to GPT-4. They are also announcing a depreciation plan for older models of the Completions AP and recommend users adopt the Chat Completions API. It comes at a time when millions of developers around the world have been requesting access to the GPT-4 AP since March.
OpenAI GPT-4 is now available to the public
GPT-4 is OpenAI’s most capable model and the products based on GPT-4 are growing everyday. From now onwards, all the existing API developers with a history of successful payments can access the GPT-4 API with 8K context. OpenAI is planning to open GPT-4 API access to the new developers by the end of this month by starting a rate limit based on computing availability.
OpenAI is also working on the stability and readiness of these models as well as in making GPT 3.5 Turbo, DALL-E, and Whisper APIs generally available. They are working on safety enabling fine-tuning on their GPT-4 and GPT 3.5 Turbo models and are expected to be available to the public later this year.
Chat Completions API which was launched in March accounts for 97% of OpenAI’s API GPT usage according to their official announcement. The Completions API which was introduced in June 2020 provides a freeform text prompt for interacting with language models. Now, the Chat Completions API has proven to be powerful with the ability to handle the vast majority of use cases and new conversational needs with higher flexibility and specificity.
To make the Chat Completions API more available and optimize compute capacity, OpenAI is going to retire some of their older models using the Completions API in 6 months. The APIs will continue to be accessible, but they will be labeled as “legacy” in OpenAI’s developer documentation starting from today. The older completion models will no longer be available starting from January 4, 2024.