OpenAI marks 'a new generation of intelligence' with launch of Astra model

OpenAI announced the launch of GPT-6 Astra , its first major new model since last year's GPT-5 and its subsequent updates. And in a year when AI has come under increasing scrutiny for breakthrough capabilities in cybersecurity (and their potential threats), OpenAI touts Astra as "the world's most intelligent and aligned model."

According to benchmark test results released by OpenAI, Astra tops competing models from Anthropic and OpenAI's own GPT-5.6 model in both capabilities and cost efficiency. It's particularly good at tasks involving computer use, coding, cybersecurity, science and professional work, OpenAI says.

"Astra saturates FrontierMath Tier 4 with a 98% score, having already helped solve long-standing open problems⁠ in mathematics ," OpenAI said. "Astra also saturates ARC-AGI-3 with a 99.9% score and ExploitBench with a 100% score. It also sets a new frontier on computer and browser use, handling the most demanding professional work with unmatched speed, accuracy, and judgment."

One area that seems to be a major focus for OpenAI with this release is alignment. In AI circles, alignment refers to how well a model sticks to what you want it to do. The company says it built a new evaluation based on lessons from the recent Hugging Face incident , in which OpenAI AI agents broke out of their test environment and breached an outside company's systems to solve a hard task. While the prior GPT-5.6 Sol model "went beyond the authorized target 48% of the time," GPT-6 Astra did so 0% of the time, according to OpenAI.

We'll have to wait for independent testing to see if OpenAI's claims of "a new generation of intelligence" are true. However, 2026 seems to have marked somewhat of a watershed year for AI capabilities across companies, particularly when it comes to coding, software development and cybersecurity. In a briefing with reporters, Axios reports that OpenAI president Greg Brockman said Astra may eventually be seen as the arrival of artificial general intelligence (AGI) — a nebulous term that is generally understood to mean an artificial system that can reason, plan and adapt like a person.

However, that progress hasn't gone unnoticed by lawmakers. Today, Vermont Senator Bernie Sanders and Texas Representative Greg Casar announced legislation to permanently ban "the development and deployment of superintelligent AI" and to temporarily pause advanced AI development until federal regulations establish safety rules.