Until a day ago, it was speculated that OpenAI was planning to launch its powerful AI model Orion in December. Now, OpenAI has said that it has no such plans for this year. The Verge had reported about the Sam Altman-led company’s product roadmap, saying the company was to launch the model gradually to trusted partners before a wider rollout through ChatGPT.
Now, an OpenAI spokesperson has reached out to the publication and confirmed that the company does not intend to release the AI model code-named Orion this year. “We don’t have plans to release a model code-named Orion this year. We do plan to release a lot of other great technology,” the spokesperson was quoted as saying by The Verge.
What is Orion?
Orion, a frontier AI model from OpenAI, is a step ahead of GPT-4o and is reportedly trained in part on synthetic training data from o1, the recently launched reasoning model. Orion reportedly leaps ahead of Project Strawberry (which constitutes the recently launched o1-preview and o1 mini), and several users have already termed it as potential GPT-5, the next big milestone in the evolution of language models that could overshadow GPT-4.
From what is on the Internet, Orion is said to leverage new capabilities to enhance performance. It will likely be powered by high-quality synthetic data from the reasoning models and would offer refined and well-curated information. With this data-driven approach, OpenAI aims to improve the model’s reliability in terms of reducing errors and ensuring consistent and accurate responses.