OpenAI launches Astra, its most powerful and controversial model to date
The company has announced the new GPT-6 Astra as the world's most powerful model, directly aiming to surpass the capabilities of its main rival, Anthropic.
OpenAI has announced its new model, GPT-6 Astra, as the most powerful in the world, aiming to surpass its main rival, Anthropic. According to the company, GPT-6 Astra is capable of performing tasks at a human level and is the most intelligent and aligned model in the world. The model is available to OpenAI clients using Daybreak, its cybersecurity program, and will be rolled out to paid plans and through Amazon Web Services over the next week.
OpenAI claims that Astra can perform various tasks, such as designing a printed circuit board, creating 3D scenes, and completing tax drafts, and has achieved a perfect score in ExploitBench, an evaluation of a model's ability to breach systems. The company has implemented safety measures, including a disalignment monitor, to mitigate potential risks associated with the model's advanced capabilities.
Written by urgent.news from Expansion ES's report — not a translation of it. Machine-written — may contain errors; check the original before relying on it.
