OpenAI announces its "next major model" Astra by dropping ten previously unsolved math solutions
OpenAI is developing "Astra," a new model family designed for multi-agent collaboration over extended periods. The company has demonstrated the system to policymakers while deciding between GPT-5 or GPT-6 branding.
OpenAI is reportedly working on a new model family codenamed "Astra." This system focuses on enabling multiple agents to collaborate on complex tasks over extended durations, ranging from hours to days.
The announcement coincides with the release of solutions to ten previously unsolved mathematics problems. CEO Sam Altman has already presented a demonstration of the technology to policymakers in Washington.
The company has not yet finalized the public naming convention for the release. Internal discussions are weighing whether to position the technology as an iteration of GPT-5 or to launch it as a distinct GPT-6 product.
This shift toward long-horizon agent collaboration marks a significant evolution in generative AI capabilities. It suggests a move beyond single-turn responses toward sustained problem-solving workflows.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.