OpenAI fully launches GPT-4! Paid API users can access immediately

Article source: Metaverse NEWS

Text: Zhou Ziyi

Currently, all OpenAI API paid users (those with successful payment history) can access GPT-4 with 8K context memory. And the company plans to open up API access to new developers at the end of this month, and then adjust the available limit through computing power availability.

OpenAI announced on Thursday (July 6) that its latest large-scale language model GPT-4 will be officially opened for use through its API.

Currently, all OpenAI API paid users (those with successful payment history) can access GPT-4 with 8K context memory. And the company plans to open up API access to new developers at the end of this month, and then adjust the available limit through computing power availability.

This means that developers around the world can use the GPT-4 large language model to enhance their own applications or develop new generative AI applications.

“Since March, millions of developers have requested access to the GPT-4 API, and the range of innovative products leveraging GPT-4 is growing every day,” OpenAI wrote in a blog post on Thursday.

In March of this year, OpenAI officially released GPT-4, which can accept image and text input to generate text (including code), and achieve “human level” on various professional and academic benchmarks, which is the same as its predecessor GPT-3.5 An improvement to the latter, which only accepts text input. Like previous GPT models, GPT-4 was trained using publicly available data, including data from public web pages as well as data licensed from OpenAI.

GPT-4 maintains the previous laurels in the context window, the default is 8k tokens (token is the basic unit for the model to process text), and the upper limit is 32k tokens. In general, models with larger context windows are better able to remember the content of recent conversations, while models with smaller windows tend to forget the above content during the conversation, causing them to deviate from the topic.

01 supports any use case

The blog post also states, “We envision a future chat-based model that can support any use case.”

OpenAI also plans to launch API interfaces for GPT-3.5 Turbo, DALL·E and Whisper.

GPT-3.5 Turbo is one of OpenAI’s other text generation models, but it is weaker than GPT-4; DALL-E 2 is OpenAI’s image generation model; and Whisper is the company’s speech-to-text model.

Image understanding capabilities are not currently available to all OpenAI customers. OpenAI is said to be testing it with a partner called Be My Eyes, but it hasn’t said when it will be available to a wider customer base.

OpenAI also said that in the future, it will allow developers to fine-tune GPT-4 and GPT-3.5 Turbo with their own data, a feature that may be available later this year.

02 Elimination and Replacement

It’s worth noting that even one of the best generative AI models available today, GPT-4, is not perfect. It misperceives facts and sometimes makes reasoning errors “with confidence” and fails on complex problems.

In this regard, OpenAI also announced that it will phase out several legacy models in the Completions API, and that future models will focus on the Chat Completions API as part of the company’s effort to optimize computing capabilities.

The picture below shows the old models that OpenAI announced to be phased out, and the new models that replaced them:

OpenAI will retire these old models of Completions API within 6 months. While the API is still accessible today, starting today the company will mark it as “legacy” in the developer documentation. Starting January 4, 2024, the old Completions API will no longer be available.