ChainThink reports that on August 1, according to The Information, citing three sources, OpenAI is preparing to launch a new model series called "Astra," focused on enhancing performance in long-duration tasks.
Reports state that OpenAI founder Altman presented Astra to policymakers and regulators in Washington this week. The series is promoted as capable of enabling multiple agents to collaborate over the long term to solve complex tasks such as projects or advanced mathematical problems.
Astra will be a new class of models following Sol, Terra, and Luna from OpenAI.
It is currently unclear when OpenAI plans to release it, or whether it will be labeled as GPT-6 or as an additional model in the GPT-5 series.
According to insiders, these models have already been tested and are expected to be among the first to be submitted under the new framework proposed by the Trump administration. The framework requires AI models to be submitted to the federal government before public release, with the government aiming to finalize the framework by the end of this weekend.