OpenAI has introduced GPT-6.2 Astra and GPT-6.2 Sol, the next generation of its flagship models, aimed at demanding professional work and everyday tasks alike. Both models launched on 17 November 2026 in ChatGPT Work, Codex and the API for Plus, Pro, Enterprise and Edu users, although they are not coming to regular Chat for now.
Astra is the new frontier model, built for the hardest projects: complex coding, scientific research, and polished professional documents. OpenAI says it follows slide templates closely, writes cleaner code, and builds better frontends, while scoring 5.2 points higher than GPT-6 Astra on Agents’ Last Exam, a benchmark for multi-step professional assignments, at less than a third of the cost per task. Sol is the efficient everyday model, and OpenAI claims it comes within half a point of Astra on Cognition’s FrontierCode 2.0 benchmark at roughly a fifth of the cost.
Pricing is part of the story. Astra costs $8 per million input tokens and $40 per million output tokens, 20% less than GPT-6 Astra, while Sol keeps its predecessor’s $2 and $10 rates. Both models offer a 1,050,000-token context window.
The launch also carries a safety note. Astra is OpenAI’s first externally released model to reach the High threshold for AI self-improvement, and both models meet the Critical threshold for cybersecurity capabilities, so OpenAI is shipping them with tighter safeguards on higher-risk security assistance. A dedicated legal configuration, Astra for Law, is available to ChatGPT Enterprise customers on request, and our guide to the best AI tools for lawyers in 2026 puts that kind of specialist tooling in context. Rival frontier systems include Claude and Gemini, which compete on the same professional workloads.
Source: OpenAI