Media: OpenAI prepares to launch a new model series, Astra, which will focus on multi-smart long-cycle task synergies

On 1 August, the Information reported that, according to three sources, OpenAI was preparing to publish a new model series, tentatively entitled “Astra”, which would be enhanced in its ability to perform long-cycle tasks. The founder of OpenAI, Altman, presented Astra to policymakers and regulators in Washington this week. It was stated that OpenAI promoted its ability to engage multiple intelligents in long-term collaborative work to address particularly difficult issues, which could be applied to project or advanced mathematical issues. Astra will be a new type of model for OpenAI after Sol, Terra and Luna. It is not clear when OpenAI plans to publish the model. OpenAI has not yet decided whether Astra will be labelled GPT 6 or as an additional model for the GPT 5 series (e.g. GPT 5.7). According to one of the sources, although the models are already being tested, they are expected to be the first to be submitted under the new framework of the Trump government plan, which requires that the AI model be submitted to the federal government before being made public. The Government has set a self-imposed deadline for this weekend, with the aim of finalizing the framework by then. In the short term, it is alleged that OpenAI also plans to issue a report on how it has addressed 10 previously unresolved mathematical issues to demonstrate its most advanced AI capabilities. Kim Xian