OpenAI has officially launched its new artificial intelligence (AI) model, GPT-5.6, approximately two weeks after it was initially shared with select agencies for safety review by the U.S. government.
According to Reuters, OpenAI announced the official release of GPT-5.6 on July 9.
GPT-5.6 consists of three models: the top-tier model 'Sol', the balanced model 'Terra', and the cost-effective model 'Luna'. Users can choose a model based on the complexity of tasks and associated costs.
This launch coincides with the U.S. government's AI safety review. GPT-5.6 was first disclosed to select agencies on June 26, in accordance with an executive order from the Trump administration regarding the pre-validation of advanced AI models. This was in response to concerns that high-performance AI could be misused for cyberattacks or sensitive technology development.
Sam Altman, CEO of OpenAI, stated in an interview with CNBC that he coordinated with U.S. government officials and incorporated various changes during that process. However, he did not disclose specific details about what was altered. He remarked that if the government review process is understandable, fair, and swift, it is acceptable.
In terms of performance, the capabilities in coding, cybersecurity, and task automation were highlighted. The GPT-5.6 Sol model slightly outperformed Anthropic's Claude Mythos 5 in terminal environment coding assessments and cybersecurity evaluations. It also scored higher than existing major models in evaluations of AI's ability to handle multiple stages of tasks independently. However, it received a lower score than Claude Mythos 5 in the SWE-bench, which measures actual software problem-solving abilities.
On the same day, OpenAI also introduced 'ChatGPT Work', a workplace AI tool based on GPT-5.6. This service is designed to assist with tasks such as document creation, coding, and data analysis. It is expected to compete in the enterprise AI market against Anthropic's Claude CoWork and Microsoft's Copilot CoWork.
In the same interview, Altman addressed reports that OpenAI proposed offering a 5% stake to the U.S. government, stating that there were many inaccuracies in those claims. He also responded uncertainly when asked about the possibility of pursuing an initial public offering (IPO) within the year.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.