OpenAI is planning to release a "materially better" GPT-5 model to replace the problematic GPT-4 sometime this year, sources told Business Insider.
Several enterprise customers noted that the new model to provide better automated services and data processing. The model's other upgrades are yet to be disclosed.
(Photo : Justin Sullivan/Getty Images)
As of writing, OpenAI is still safety testing the new GPT model with only a select few companies able to access the demo.
The report came in after OpenAI CEO Sam Altman said the GPT-4 model "kinda sucks." The AI firm's current chatbot model has been reported several times of having issues that render it unusable.
Just last month, many users complained that ChatGPT seems to be suffering from "stroke" after generating incomprehensible responses. Before that, the chatbot was accused of getting "lazy."
Also Read: OpenAI CEO Sam Altman is Not Impressed with GPT-4
OpenAI CEO Sam Altman Moves On from GPT-4, GPT-3
Rumors of the upcoming GPT-5 model has been going around for quite some time as the company ramps updates on its GPT-4.5 and GPT-4.5 Turbo, while slowly deprecating the GPT-3 and the base GPT-4 model.
In one of his recent appearances, Altman even called the older models to be "unimaginably horrible" in comparison to the company's newest development.
It is worth noting, however, that Altman referred to GPT-4 as a "brainstorm partner" to inspire them in building its successor.
Altman is yet to disclose much detail about the GPT-5 but promised customers to expect "many different things" in the coming months, Readwrite reported.
Related Article: ChatGPT is Getting 'Lazy' Which Could Have Been Learned from Training Data
OpenAI Brings More Content to ChatGPT
Ahead of the release of its GPT-5 model, OpenAI has already been expanding its training data to better improve its AI models in, at least, legal ways.
The AI firm has recently been reported to finishing several deals with news publishers to acquire their licensed content for AI training in exchange for offering their services on news platforms.
Many of the deals signed involved billions of dollars.
The move comes as OpenAI continues to evade copyright lawsuits from authors and publishers, including the New York Times, for using their licensed content without permission or monetary compensations.
