FoxTPNL / Wikimedia Commons (CC BY 4.0) The new model solved ten open math problems for $2,000 in compute and hit a cybersecurity threshold that no previous AI system has reached
OpenAI’s newest model family, known as Astra, is doing things that would have sounded like science fiction two years ago. It cracked ten longstanding open problems in mathematics and theoretical computer science, producing verified proofs at a compute cost of roughly $2,000. For context, some of those problems had stumped human researchers for decades.
First publicly named on August 1, 2026, Astra represents a substantial leap beyond the GPT-5.6 family that OpenAI released on July 9, 2026. Whether the company officially brands it as GPT-6 remains an open question, but the capabilities gap between Astra and its predecessor is wide enough that the label feels earned.
What Astra can actually do #
The math results are impressive on their own, but the cybersecurity milestone might matter more in practical terms. On September 1, 2026, Astra became the first model to meet the “Critical” cybersecurity capability threshold under OpenAI’s Preparedness Framework. That means it can autonomously discover and exploit previously unknown vulnerabilities.
OpenAI is keeping that particular feature locked down. The advanced cyber capabilities are restricted to a small group of testers at launch.
On the commercial side, early results look promising. A case study published on September 3, 2026 showed that Playco, a gaming company, experienced 50% fewer manual fixes when using Astra for game prototyping.
The AGI question #
Sam Altman and OpenAI’s leadership have stated they expect to achieve an internal system that could be termed AGI by the end of 2026. Astra is positioned as a critical step in that direction, particularly through its ability to automate research tasks that previously required human experts.
OpenAI has been building toward this moment with deliberate pacing. The company released GPT-5.6 in a limited preview before its broader July launch, then began private demonstrations of Astra’s capabilities to US policymakers.
What this means for the market #
The productivity gains demonstrated by Playco offer a preview of how Astra could reshape enterprise software economics. If a 50% reduction in manual fixes scales across other use cases, companies that integrate Astra into their workflows early stand to gain a meaningful competitive advantage.
The cybersecurity implications cut both ways. Organizations using Astra defensively could identify and patch vulnerabilities before human attackers find them. But the existence of a model capable of autonomous exploit discovery also raises the stakes for every company that doesn’t have access to it.
The $2,000 compute cost for solving those ten math problems is also worth noting as a business signal. If frontier-level research can be conducted at that price point, the economics of AI-assisted R&D shift dramatically. Academic institutions and smaller companies that previously couldn’t afford to run large-scale AI experiments may suddenly find themselves with access to capabilities that were exclusive to well-funded labs just months ago.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our