According to CoinMeta, OpenAI announced that its new GPT-6 Astra has surpassed Anthropic to become the world's most intelligent AI model. The company stated that Astra is in a leading position in the fields of software engineering, science, and network security, but did not release any standard benchmark comparisons. Anthropic's Claude Fable currently leads the existing models of OpenAI in the rankings of Benchlm and Swe-bench. OpenAI shows that Astra scored 100% in network attack benchmark tests and solved 10 mathematical problems that have remained unsolved for decades. OpenAI's own security team rated this model as "critical," which is its highest level of danger. President Brockman indicated that Astra could be AGI, but claimed that AGI is more like a conceptual idea. The billion-dollar deals with Microsoft and Amazon include AGI clauses, and if AGI is officially announced, it could reshape these partnerships.