Source : INDIA TODAY NEWS
OpenAI has introduced three new artificial intelligence models named Sol, Terra, and Luna, but their release has not followed the usual global rollout. The company said the models will first be available only to a small set of trusted partners after the US government asked for a limited launch instead of open access.
The announcement came through a post by OpenAI CEO Sam Altman on X, where he described the new systems as a major step forward in capability and pricing efficiency. However, he also confirmed that users will not get full access right away due to regulatory requirements tied to early safety review of advanced AI models.
advertisement
Altman said Sol is a smart and efficient model and a clear upgrade in capability. It is priced at the same level as GPT-5.5 but offers improved performance in several areas. Along with it, OpenAI introduced Terra, which is part of the GPT-5.6 family and delivers performance similar to GPT-5.5 but at half the cost.
A third model, Luna, has also been introduced under the same lineup. OpenAI has grouped these models under GPT-5.6, positioning them as part of its next step in model development, with a focus on better performance and more cost-efficient access for developers and enterprises.
Altman also responded to a user question asking whether the models would be available globally, saying, “working hard for worldwide”
US government request leads to limited rollout plan
OpenAI said the models are currently being released only in a restricted preview for selected partners because of a request from the US government. The company added that it had already shared information about the models with authorities before launch as part of early evaluation efforts.
In its official note, OpenAI said it supports wider access to its systems and is working to make the models available to more users in the coming weeks. At the same time, it added that the current access rules should not become a long-term practice, as it can delay access for developers, enterprises, and security teams who depend on such tools.
“We don’t believe this kind of government access process should become the long-term default,” the company said.
OpenAI also stated that it is working with the US administration to build a more structured system for reviewing advanced AI models before release. The aim is to create a repeatable process for future launches that balances safety checks with faster availability.
The development comes as AI companies in the US face closer attention from regulators. The Trump administration has recently taken a more active approach to AI oversight, asking companies to allow early review of powerful models before public release.
OpenAI said it expects Sol, Terra, and Luna to reach general availability in the coming weeks, but has not shared a fixed timeline or confirmed which partners currently have access.
The company also said Sol shows improvements in areas such as coding and biology, and is its strongest model so far for cybersecurity-related tasks. It is better at helping fix security issues in software, while still staying within OpenAI’s internal safety limits for high-risk behaviour.
– Ends
SOURCE :- TIMES OF INDIA
