OpenAI Launches GPT-6 Astra as Company Claims AI Has Entered the AGI Era

OpenAI has officially launched GPT-6 Astra, presenting its newest flagship model as a major step toward artificial general intelligence (AGI). Released on September 3, Astra is designed to perform a wide range of complex tasks involving computers, software development, professional work, scientific research and web-based activities. OpenAI says the model represents a significant improvement over its previous systems and is its most capable model yet.

One of Astra’s biggest improvements is its ability to interact with computers rather than simply responding to text prompts. The model can reportedly complete tasks such as searching for services, finding job opportunities, creating software and handling complicated professional workflows. OpenAI demonstrated that Astra could complete some research-heavy tasks considerably faster than a human working through the same process, highlighting the company’s broader push toward AI agents capable of carrying out multi-step assignments independently.

OpenAI is also highlighting Astra’s performance on demanding AI benchmarks. The company reports a 99.9% score on ARC-AGI-3, along with strong results in advanced mathematics, coding and cybersecurity evaluations. These results are being used by OpenAI to support its argument that Astra represents a new level of general-purpose AI capability. However, benchmark scores remain only one way of measuring intelligence, and independent testing will be important in determining how well those results translate into everyday real-world performance.

The launch has also attracted attention because Astra has reached OpenAI’s highest internal cybersecurity risk classification, known as “Critical.” The designation reflects the model’s ability to identify vulnerabilities and perform sophisticated cybersecurity-related tasks. OpenAI says it has introduced additional monitoring and safety measures around the model, while emphasizing defensive applications. The company is also launching Daybreak for frontline defenders, backed by a $1 billion commitment intended to help organizations strengthen cybersecurity and protect essential services and critical infrastructure.

Astra’s arrival therefore represents both an important technological milestone and a major safety test for OpenAI. The company is positioning the model as evidence that increasingly capable AI can contribute to scientific discovery, software development, professional work and cybersecurity. At the same time, its advanced capabilities mean that monitoring, alignment and responsible deployment will become increasingly important. OpenAI President Greg Brockman described the release as the beginning of the “AGI era,” but how history ultimately judges that claim will depend on Astra’s real-world performance, reliability and safety.