OpenAI is reportedly working on fresh strategies to address a noticeable slowdown in the rate of improvement with its AI models. According to The Information, early testers of OpenAI’s next anticipated model, known internally as “Orion,” observed that while Orion does outperform the company’s existing models, the progress is less dramatic than the leap from GPT-3 to GPT-4.
This slower rate of advancement suggests that OpenAI may be approaching a plateau in certain areas, like coding capabilities, where Orion doesn’t consistently surpass its predecessors.
To counter this, OpenAI has established a new “foundations team” tasked with exploring how to advance its models, even as the availability of fresh training data becomes scarce. This group is reportedly investigating innovative techniques, including using synthetic data generated by other AI models and enhancing model performance through advanced post-training adjustments.
Although OpenAI did not comment on this report directly, it previously clarified that it had “no plans to release a model called Orion this year.”
We have helped 20+ companies in industries like Finance, Transportation, Health, Tourism, Events, Education, Sports.
