Today, OpenAI made GPT-4, their latest text-generating model, API-available.
All OpenAI API developers “with a history of successful payments” can access GPT-4 this afternoon. By the end of the month, the company will let additional developers and then raise availability limits “depending on compute availability.”
“Millions of developers have requested access to the GPT-4 API since March, and the range of innovative products leveraging GPT-4 is growing every day,” OpenAI stated in a blog post. “We envision chat-based models supporting any use case.”
GPT-4 can generate text (including code) and receive image and text inputs, improving on GPT-3.5, which only accepted text, and performs at “human level” on several professional and academic benchmarks. GPT-4, like OpenAI’s earlier GPT models, was trained using public web pages and licensed data.
OpenAI customers cannot currently use image-understanding. OpenAI is testing it with one partner, Be My Eyes. It hasn’t said when it’ll open it to more customers.
GPT-4, like many generative AI models, has flaws. It “hallucinates” facts and mistakes reasoning, sometimes confidently. It fails to learn from its mistakes, injecting security flaws into code it generates.
OpenAI will soon allow developers to fine-tune GPT-4 and GPT-3.5 Turbo, one of its other recent but less capable text-generating models (and one of the original models powering ChatGPT), with their own data. OpenAI expects it later this year.
Generative AI rivalry has intensified since GPT-4’s March announcement. Anthropic increased Claude’s context window from 9,000 to 100,000 tokens. (Context window refers to the text the model considers before generating more text, while tokens reflect raw text—e.g., “fantastic” would be split into “fan,” “tas,” and “tic.”)
GPT-4 was the previous context window champion at 32,000 tokens. Models with small context windows sometimes “forget” even recent interactions, veering off subject.
Today, OpenAI released its DALL-E 2 and Whisper APIs, which are the company’s image-generating and speech-to-text models, respectively. The company aims to deprecate obsolete API models to “optimize [its] compute capacity.” OpenAI has struggled to meet demand for its generative models in recent months due to ChatGPT’s popularity.
GPT-3 and its derivatives will be replaced with new, more compute-efficient “base GPT-3” models on January 4, 2024. Developers utilizing the old models must manually upgrade their integrations by January 4, and those who want to continue using fine-tuned old models must fine-tune replacements atop the new base GPT-3 models.
“We will support users who previously fine-tuned models to make this transition as smooth as possible,” OpenAI noted. We will contact developers who recently used these older models in the coming weeks and provide more information once the new completion models are ready for early testing.