ChainThink report, August 1, according to three informed sources cited by The Information, OpenAI is preparing to launch a new model series called "Astra," focused on enhancing long-horizon task completion capabilities.

The report states that OpenAI founder Altman has already demonstrated Astra to policymakers and regulators in Washington this week. The series is promoted as enabling multiple agents to work collaboratively over extended periods, tackling complex tasks such as project execution or advanced mathematical problems.

Astra will become OpenAI's newest class of models following Sol, Terra, and Luna.

It remains unclear when OpenAI plans to release the model, nor has it decided whether it will be labeled GPT-6 or introduced as an additional model within the GPT-5 series.

According to informed sources, these models are already under testing and are expected to be among the first to be submitted under the new framework proposed by the Trump administration. The framework mandates that AI models be submitted to the federal government prior to public release, with the government aiming to finalize the framework by the end of this weekend.