
OpenAI
On Thursday, OpenAI announced that all paying API customers now have access to the GPT-4 API. It also introduced updates to the chat-based models, announced the transition from the Completions API to the Chat Completions API, and outlined plans for the removal of old models.
Generally considered the most powerful API product, the GPT-4 API was first launched in March but has been under closed testing until recently. As an API, developers can use a special interface to integrate OpenAI’s large language model (LLM) into their own products for uses such as summarization, coding assistance, analysis, and composition. The model runs remotely on OpenAI servers and provides output to other apps on the Internet.
OpenAI says the GPT-4 API with 8K context will be accessible to developers with a successful payment history, with plans to open access to new developers by the end of July . And in a move to distance itself from the old GPT-3-style models, OpenAI has also chosen to begin retiring the “Completions API” models in favor of the newer Chat Completions API models. Since its launch in March, OpenAI says that Chat Completions API models now account for 97 percent of OpenAI’s API GPT usage.
This depreciation plan includes retiring models that are part of the Completions API for six months. Starting January 4, 2024, these older models, which only complete a prompt instead of engaging in a chat-like dialog, will be replaced by newer versions, and developers using some of them models are required to upgrade their integration manually. For now, the old models will remain available but will be labeled as “legacy” products. Here is a partial list of discontinued models:
- fairy
- babbage
- curious
- davinci
- davinci-instruct-beta
- curie-instruct-beta
- text-ada-001
- text-babbage-001
- text-curie-001
- text-davinci-001
- text-davinci-002
- text-davinci-003
OpenAI will provide drop-in replacements for these older models. For example, OpenAI recommends that users of text-davinci-003 model switch to gpt-3.5-turbo-instruct model instead. Similarly, the ada-002, babbage-002, curie-002, and davinci-002 models replace the previous versions of each. Applications that use some of these GPT-3 models (such as ada, babbage, curie, davinci) “will automatically be upgraded to the new models listed above on January 4, 2024,” according to OpenAI.
OpenAI also announced that “based on the stability and readiness of these models for production scale use,” it is also making APIs for Whisper, DALL-E, and GPT-3.5 Turbo “commonly available. ” And the company expects to continue to improve the models throughout the year.
Developers can find more details in OpenAI’s blog post for the announcement.