OpenAI releasing major upgrade to ChatGPT and Codex with GPT-6 Astra, details here

OpenAI has unveiled a groundbreaking upgrade to its ChatGPT and Codex platforms, introducing GPT-6 Astra - touted as the world's most intelligent and aligned model. The new model boasts state-of-the-art capabilities across a range of applications including computer use, browsing, software engineering, cybersecurity, science, and professional work.

In computer use, GPT-6 Astra reaches the FrontierMath Tier 4 with a 98% score, solving longstanding mathematical problems and handling demanding tasks with unmatched speed and accuracy. It also excels in ARC-AGI-3 achieving a 99.9% score and perfecting ExploitBench at 100%.

For software engineering, GPT-6 Astra is considered the best model to date. It combines the intelligence required for complex problems with the ability to execute multistep workflows and produce polished documents, spreadsheets, and presentations.

A significant improvement for Codex is its ability to preserve and retrieve context even when the context window fills. Traditionally, models would compress work during long sessions, potentially losing critical details. However, Astra can maintain notes across context windows, preserving accumulated details without repeatedly compressing them.

GPT-6 Astra is currently being released to a limited set of organizations with plans to roll out to all ChatGPT Plus, Pro, Business, and Enterprise users, as well as through the OpenAI API and AWS. The new model is included in existing subscription allowances, with additional usage available for purchase. Enterprise administrators have the option to enable Astra for their workspace post-launch.

While GPT-5 launched in August 2025, replacing GPT-4o and other reasoning models, GPT-6 Astra represents a significant leap forward in OpenAI's commitment to creating the most intelligent and aligned AI models.