GPT-4 by OpenAI Now Available to the Public
OpenAI Releases GPT-4, its Latest Text-Generating Model, to Developers
chandni
OpenAI has officially launched GPT-4, its newest text-generation model, making it accessible to developers through the OpenAI API. Existing API developers with a proven payment history gained immediate access to GPT-4, while OpenAI plans to extend access to new developers by the end of the month, gradually increasing availability limits based on computational resources.
Since March, OpenAI has received millions of requests for access to the GPT-4 API, demonstrating a growing interest in leveraging the model for innovative products. OpenAI envisions a future where chat-based models can cater to a wide range of use cases.
Compared to its predecessor GPT-3.5, GPT-4 brings significant improvements. It has the ability to generate text, including code, and accepts both image and text inputs. OpenAI claims that GPT-4 performs at a "human level" on various professional and academic benchmarks. The model was trained using publicly available data, including web pages, as well as licensed data.
However, the image-understanding capability of GPT-4 is currently limited to testing with a single partner, Be My Eyes. OpenAI has not specified when this feature will be available to a broader customer base.
It's important to note that GPT-4, like other generative AI models, is not flawless. It may produce "hallucinated" facts and make reasoning errors, sometimes with high confidence. Additionally, it does not learn from experience and struggles with challenging problems like introducing security vulnerabilities into generated code.
OpenAI plans to allow developers to fine-tune both GPT-4 and GPT-3.5 Turbo, another recent text-generating model, using their own data. This capability, currently available for some of OpenAI's other models, is expected to be implemented later this year.
The competition in generative AI has intensified since the introduction of GPT-4 in March. Anthropic, for example, has expanded the context window of its text-generating model, Claude, from 9,000 tokens to 100,000 tokens. GPT-4 previously held the crown for the largest context window at 32,000 tokens. Models with smaller context windows tend to forget the content of recent conversations, leading to deviations from the intended topic.
In a related announcement, OpenAI has made its DALL-E 2 (image-generating model) and Whisper (speech-to-text model) APIs generally available. To optimize computational capacity, OpenAI plans to deprecate older models offered through its API. Starting from January 4, 2024, models such as GPT-3 and its derivatives will be replaced with new "base GPT-3" models, which are presumably more computationally efficient. Developers using the old models will need to manually upgrade their integrations, while those wishing to continue using fine-tuned old models must fine-tune replacements based on the new base GPT-3 models.
