OpenAI Rolls Out GPT-6 Astra, Its ‘Most Capable Model’ Yet

OpenAI has unveiled GPT-6 Astra, its most advanced AI model to date, designed to handle complex tasks autonomously. This frontier model boasts a record-breaking 1.05 Mn-token context window and can generate outputs up to 1.28 lakh tokens. Capable of autonomously navigating websites, interacting with software, working with documents, writing and testing code, and managing multi-step workflows, Astra sets a new standard for AI systems.

The model has scored highly on several benchmarks, including FrontierMath Tier 4, ARC-AGI-3, and ExploitBench, indicating its proficiency in reasoning tasks. It is equipped with tools such as web search, file search, computer use, code execution, and image generation. Initially, GPT-6 Astra will be accessible to a select group of organizations before being made available to all ChatGPT Plus, Pro, Business, and Enterprise users.

Priced at $10 per Mn input tokens and $50 per Mn output tokens, with discounted rates for cached inputs, the model is positioned as a significant leap forward for AI interfaces, particularly in coding, science, cybersecurity, and professional work. OpenAI CEO Sam Altman highlighted its superior performance in computer use and professional tasks, positioning it as the best model in the world for these applications.

GPT-6 Astra's enhanced capabilities in coding, scientific and mathematical applications, and cybersecurity underscore its versatility and mark a significant step in the ongoing race among AI companies to develop increasingly capable AI agents.