Starting this afternoon, all existing developers of the OpenAI API with a satisfactory payment history can access GPT-4.
The company plans to open access to new developers by the end of this month and then gradually increase availability limits based on computing availability.
GPT-4 has the ability to generate text, including code, and accept both image and text inputs, marking an improvement over its predecessor, GPT-3.5, which only accepted text inputs. It has demonstrated “human-level” performance in various professional and academic benchmark tests.
Similar to previous GPT models from OpenAI, GPT-4 was trained using publicly available data, including data from public web pages, as well as licensed data owned by OpenAI.
OpenAI’s challenges in the coming months with GPT-4
The image comprehension feature is not yet available to all OpenAI customers. Initially, OpenAI is testing it with a single partner, Be My Eyes. However, they have not indicated when it will be opened up to a wider customer base.
It’s worth noting that, like other current state-of-the-art generative AI models, GPT-4 is not perfect. It can be “gullible” with facts and may make reasoning errors. It also does not learn from experience, failing in challenging problems such as introducing security vulnerabilities in the generated code.
In the future, OpenAI states that they will allow developers to fine-tune GPT-4 with their own data, as has been possible with other OpenAI text generation models for some time. According to OpenAI, this feature is expected to be available by the end of the year.
Since the introduction of GPT-4 in March, the competition in the generative AI space has intensified. Recently, Anthropic expanded the context window of Claude – their flagship text-generating AI model, still in preview – from 9,000 to 100,000 tokens.
GPT-4 retained its previous crown in terms of context window, with 32,000 tokens at the upper end. In general, models with smaller context windows tend to “forget” the content even of very recent conversations, leading to topic drift.
Starting from January 4, 2024, some older OpenAI models – specifically GPT-3 and its derivatives – will no longer be available and will be replaced by new “base GPT-3” models, which are expected to be more computationally efficient.
Some of the links added in the article are part of affiliate campaigns and may represent benefits for Softonic.
