{"slug": "openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement", "title": "OpenAI says GPT-6.2 Astra is first deployed model to hit High self-improvement threshold", "summary": "OpenAI released GPT-6.2 Astra on November 17, 2026, the first externally deployed OpenAI model to meet the High threshold for AI Self-Improvement under the company's Preparedness Framework, alongside sibling model GPT-6.2 Sol, which reached Critical for cybersecurity and High for biological and chemical risks. Both models are live in ChatGPT Work, Codex and the API, with GPT-6.2 Astra priced at $8/$40 per million tokens, less than its September 2026 predecessor GPT-6 Astra, and improving 5.2 points on Agents' Last Exam. The classifications are self-reported under OpenAI's own framework rather than an outside audit.", "body_md": "# OpenAI says GPT-6.2 Astra is first deployed model to hit High self-improvement threshold\n\nFoxTPNL / Wikimedia Commons (CC BY 4.0)\n\nOpenAI's new flagship crossed a line in its own Preparedness Framework, while sibling model GPT-6.2 Sol landed at Critical for cybersecurity\n\n[OpenAI](https://cryptobriefing.com/markets/openai/) has released a model that, by its own measurement, is meaningfully good at helping build better AI.\n\nGPT-6.2 Astra is the first externally deployed model to meet the **High threshold for AI Self-Improvement** under OpenAI’s Preparedness Framework. The company built that scorecard to track exactly this kind of capability.\n\nThe model arrived on November 17, 2026, alongside a sibling called GPT-6.2 Sol. Both are now live in ChatGPT Work, in Codex and through the API.\n\n## Two models, two very different risk report cards\n\nOpenAI is positioning the pair as a split lineup. Astra is the heavier, more capable option. Sol is the cost-effective model meant for everyday use.\n\nBoth are built for professional workloads, and both are described as more efficient than what came before.\n\nThe safety classifications are where things get interesting. OpenAI’s Preparedness Framework sorts dangerous capabilities into categories and grades each one against tiers. The labels named in this release are High and Critical.\n\nAstra is the headliner because it cleared the High bar for self-improvement. No externally deployed OpenAI model had done that before.\n\nSol did not reach High on self-improvement. Sol meets the **Critical** threshold for cybersecurity and the High threshold for biological and chemical risks.\n\n### AI, tech, and the markets they move—in one daily briefing.\n\nDaily. Free. Join 34,000+ readers across crypto, finance, and policy.\n\n## How Astra stacks up against its predecessor\n\nAstra replaces GPT-6 Astra, which OpenAI released in September 2026. That earlier model posted a perfect score on ExploitBench. Even so, it fell short of the High self-improvement threshold.\n\nRoughly two months later, the successor clears that bar. On the benchmarking side, GPT-6.2 Astra improves on GPT-6 Astra by 5.2 points on Agents’ Last Exam.\n\nOpenAI also cut the price. GPT-6.2 Astra costs less than GPT-6 Astra, with general-user pricing set at $8/$40 per million tokens.\n\n## Why the self-improvement category matters\n\nMost AI risk categories describe what a model could help a person do. Think of assisting with a cyberattack or providing dangerous chemistry knowledge.\n\nSelf-improvement is a different kind of concern. It asks whether a model can contribute to the development of AI itself, potentially speeding up its own successors.\n\nThat is the idea behind the phrase recursive self-improvement. OpenAI says it is continuing to evaluate that capability as part of its ongoing safety work.\n\n## What this means for developers, businesses and the industry\n\nFor developers and enterprise buyers, the immediate story is practical. A more capable flagship is available at a lower price than its predecessor. A cheaper everyday option ships alongside it.\n\nThat two-tier structure lets teams route demanding work to Astra and routine tasks to Sol.\n\nDistribution matters too. Shipping both models through ChatGPT Work, Codex and the API at once puts them in front of business users, coders and developers on day one.\n\nThe safety picture is harder to wave through. OpenAI is deploying a model at High for self-improvement and a model at Critical for cybersecurity in the same release.\n\nThose classifications come from OpenAI’s own framework. That makes them a self-reported warning label, not an outside audit.\n\n**Disclosure:** This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement", "canonical_source": "https://cryptobriefing.com/openai-gpt-6-2-astra-self-improvement-threshold/", "published_at": "2026-11-17 18:53:13+00:00", "updated_at": "2026-11-17 20:28:53.449315+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "large-language-models", "ai-policy", "ai-products"], "entities": ["OpenAI", "GPT-6.2 Astra", "GPT-6.2 Sol", "GPT-6 Astra", "Preparedness Framework", "ChatGPT Work", "Codex", "Agents' Last Exam"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement", "markdown": "https://wpnews.pro/news/openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement.md", "text": "https://wpnews.pro/news/openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement.txt", "jsonld": "https://wpnews.pro/news/openai-says-gpt-6-2-astra-is-first-deployed-model-to-hit-high-self-improvement.jsonld"}}