OpenAI President Greg Brockman used a September 3 company briefing to formally mark what he called the arrival of the 'AGI era,' anchoring the claim to the rollout of the company's new GPT-6 Astra system. The declaration, delivered at the end of the session, signals that OpenAI's leadership now wants the GPT-6 line treated as a generational leap rather than an incremental upgrade, with Astra positioned as the flagship interface for that push.
The 'AGI era' framing matters because it gives OpenAI a marketing and strategic anchor at a moment when frontier-lab rivals are closing in on benchmark parity. By naming the period after its own product, the company is attempting to set the narrative clock for the industry, much as search engines, smartphones and cloud computing were each branded as the next defining compute era.
ALSO READ | Apple’s First Foldable iPhone Duo Set for Tonight’s Launch with Leaked Color Lineup and Pricing
Anthropic's contribution to the same news cycle was more numerical and arguably more concrete. Its Fable 5.1 model more than doubled the science score recorded by its predecessor, according to figures circulating after the briefing. A doubling on a science-focused benchmark is a serious jump in capability, particularly for tasks involving research reasoning, structured problem solving and technical analysis, which are increasingly used by enterprises evaluating which model to deploy.
The competitive picture still tilts toward two cloud partners. Microsoft, the long-standing backer of OpenAI, and Amazon, a major investor and infrastructure partner for Anthropic, both benefit from the latest gains, though analysts tracking the disclosures suggest one of the two holds a slight edge going into the next product cycle. The exact margin was not detailed in the briefing materials, but the framing points to a tightening race in which model quality and cloud distribution are moving in lockstep.
Taken together, the announcements underline how quickly the frontier-model race is being reframed around science-style reasoning scores and AGI-style rhetoric rather than raw chat benchmarks. For enterprise buyers, the OpenAI-Anthropic duel now offers two credible stacks — one wrapped in Microsoft's Azure and enterprise sales motion, the other increasingly woven into Amazon Web Services' Bedrock and chips roadmap.
