OpenAI launches GPT-6 Astra with advanced benchmark claims

OpenAI announced on X the release of GPT-6 Astra, its new flagship model, claiming state-of-the-art performance on three core benchmarks: FrontierMath Tier 4, ARC-AGI 3, and TerminalBench-4.0. The announcement included no methodological detail or numerical comparisons with prior models.

Beyond the general benchmarks, the company pointed to significant progress in scientific discovery, citing leading results on Terminal-Bench Science 0.1 and HealthBench Pro. Those tests evaluate the ability to work in scientific terminal environments and on complex health tasks, respectively, but the source did not disclose exact scores or comparison baselines.

The model begins rolling out today to a limited group of organizations and will open to all subscribers on the ChatGPT Plus, Pro, Business, and Enterprise tiers, as well as through the OpenAI API and AWS, over the coming days. The company provided no precise schedule beyond "the coming days" and did not detail usage limits or separate pricing for Astra.

The announcement comes directly from the manufacturer without an accompanying technical report, peer-reviewed paper, or independent test results. In the absence of transparency around measurement methodology, sample size, and run conditions, it is difficult to assess the practical significance of the claims beyond the marketing statement.