GPT-6 Astra: A new generation of intelligence

OpenAI shipped GPT-6 Astra as its new flagship model, calling it the most capable and aligned system it has released, with explicit focus on computer use, software engineering, math and science, office work, and cybersecurity. The rollout starts with select organizations before reaching ChatGPT tiers, the API, and AWS. Independent testing already shows why the launch matters beyond marketing: Simon Willison found Astra pricing matches Claude Fable at $10/$50 per million input/output tokens, while the model posts a 99.9% ARC-AGI-3 score under OpenAI's custom harness and near-100% recall on an 8-needle test out to 1M tokens. Latent Space's hands-on review, after burning more than 20 billion tokens, reports Astra can independently choose and train models, label data for active learning, monitor pipelines and logs, debug deployed systems, and orchestrate subagents across long-running threads. OpenAI's own safety overview classifies Astra as the first broadly deployed model to reach the Critical level of cybersecurity capability under its Preparedness Framework, a formal designation with direct consequences for access controls. Taken together, Astra resets the reference point agentic engineering teams should benchmark against, at a price point they are already used to paying.