Back to Flash News

OpenAI launches GPT-6.2 Astra and GPT-6.2 Sol

OpenAI announced on November 17, 2026, that as of today, GPT-6.2 Astra and GPT-6.2 Sol are officially available to Plus, Pro, Enterprise and Edu users, and can be used in ChatGPT Work and Codex.

According to OpenAI’s announcement: Pro 500 subscribers can enable Ultrafast mode for both models; Pro 200 users are receiving Ultrafast for the first time, and for now it only supports GPT-6.2 Sol. Developers can call gpt-6.2-astra and gpt-6.2-sol through the OpenAI API and use Ultrafast. Eligible Enterprise and Edu workspaces can get the same access after an administrator enables it.

OpenAI describes GPT-6.2 Astra as a frontier model for the hardest projects, setting new highs on professional work and scientific research evaluations, with per-task costs less than one-third of GPT-6 Astra’s; GPT-6.2 Sol is a high-efficiency general-purpose model for everyday work. Because of the improved capabilities, the two models adopted stricter safety measures at launch for high-risk cybersecurity assistance.

In terms of API pricing, GPT-6.2 Astra is priced at $8 per million input tokens and $40 per million output tokens; GPT-6.2 Sol is priced at $2 input and $10 output.

Original source:
OpenAI Developer Community announcement, GPT-6.2 Astra and Sol are now available (November 17, 2026), with content consistent with the description of OpenAI’s official YouTube video and @OpenAI’s official post.
Link: community.openai.com/t/gpt-6-2-astra-and-sol-are-now-available/1417531

ABAB AI Insight

  1. Product and positioning background: OpenAI continues its dual-model strategy of Astra (high-capability frontier) and Sol (efficient for everyday use). Astra focuses on complex professional tasks, scientific research and difficult coding, while Sol emphasizes multi-step agent workflows and cost efficiency; many OpenAI employees use it as the default model.

  2. Access and commercial logic: Ultrafast (high-speed generation) is provided through differentiated subscription tiers and is being opened to Pro 200 for the first time, helping broaden coverage among mid- to high-tier paying users; the API launch at the same time makes it easier for developers to integrate, and Astra’s pricing is down about 20% from the previous generation.

  3. Safety and launch context: The improved capabilities, especially on cybersecurity evaluations, are accompanied by tighter default protections, reflecting OpenAI’s cautious approach when deploying high-capability models; earlier related versions were adjusted for safety reasons.

  4. Competition and ecosystem comparison: The launch timing comes amid competition with rivals such as Anthropic on evaluation leaderboards; partners such as GitHub Copilot are also going live at the same time, expanding real-world use cases.

Cognitive law:
A leap in model capability is often accompanied by layered release of access rights and simultaneous tightening of safety constraints — companies expand their user base through “high capability + differentiated speed + API openness,” while managing the spillover of high-risk capabilities with protective measures.