OpenAI Launches GPT-6 Astra; Claims Lead Over Claude Fable 5.1 In Coding & Reasoning

OpenAI has introduced GPT-6 Astra, its latest flagship model, touting remarkable advancements in computer usage, coding, cybersecurity, and scientific reasoning. The new model is positioned as a significant leap over Anthropic's Claude Fable 5.1, excelling in various benchmarks and applications.

Rollout details indicate that GPT-6 Astra is initially available to a limited group of organizations before becoming accessible to all ChatGPT Plus, Pro, Business, and Enterprise users, along with OpenAI API, Microsoft Azure, and AWS Bedrock users. Subscription allowances will cover GPT-6 Astra usage, with an option for additional credits. Companies enrolled in OpenAI's application-based cybersecurity program gain early access.

Pricing on the API is set at $10 per million input tokens and $50 per million output tokens, with lower rates for cached input and a "Fast" processing mode priced at roughly double the standard rate for approximately twice the speed. The model offers a context window of 1,050,000 tokens and a maximum output of 128,000 tokens, roughly 2.5 times the promotional pricing of OpenAI's previous model, GPT-5.6 Sol.

GPT-6 Astra is described as the most intelligent and aligned model to date, showcasing state-of-the-art performance in computer usage, browsing, software engineering, cybersecurity, science, and professional work. The model significantly outperforms its predecessor on various benchmarks, achieving near-perfect scores in advanced mathematics reasoning and simulated computer environments while completing everyday computer tasks 47% faster than GPT-5.6 Sol.

GPT-6 Astra demonstrates impressive gains in producing business documents, spreadsheets, and slide decks that match user templates and styles, and it asks clarifying questions only when decisions genuinely depend on the answers. In coding, Astra is considered the best model for software engineering, with improved agentic coding performance and reduced iteration on generated code.

The model also boasts improved cybersecurity capabilities, scoring perfectly in vulnerability exploitation tests and independently finding two previously unknown security flaws.

When compared to Claude Fable 5.1, OpenAI's benchmark tables show Astra outperforming the latter in several categories, including coding, professional work, science, cybersecurity, and abstract reasoning. Notably, Astra achieves perfect or near-perfect scores in key reasoning benchmarks, surpassing both GPT-5.6 Sol and Claude Fable 5.